{
  "source_base": "https://echo.unipat.ai",
  "captured_at": "2026-07-01",
  "endpoints": {
    "rankings_overall": {
      "path": "/api/v2/rankings?category=Overall",
      "status_code": 200,
      "data": {
        "success": true,
        "data": [
          {
            "rank": 1,
            "modelId": 29,
            "modelName": "EchoZ-1.0",
            "provider": "official",
            "description": "",
            "organization": "UniPat AI",
            "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
            "eloScore": 1024.1,
            "battles": 83606,
            "eloMargin": 1.3,
            "rankUp": 1,
            "rankDown": 1,
            "resolvedCount": 2396,
            "firstPredictionDate": "03-04"
          },
          {
            "rank": 2,
            "modelId": 56,
            "modelName": "Kimi-K2.6",
            "provider": "official",
            "description": "",
            "organization": "Moonshot AI",
            "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
            "eloScore": 1011.2,
            "battles": 15258,
            "eloMargin": 3.2,
            "rankUp": 2,
            "rankDown": 4,
            "resolvedCount": 672,
            "firstPredictionDate": "04-24"
          },
          {
            "rank": 3,
            "modelId": 46,
            "modelName": "Market",
            "provider": "system",
            "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
            "organization": "Polymarket",
            "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
            "eloScore": 1011.0,
            "battles": 74191,
            "eloMargin": 1.3,
            "rankUp": 2,
            "rankDown": 4,
            "resolvedCount": 1759,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 4,
            "modelId": 58,
            "modelName": "Deepseek-V4-Pro",
            "provider": "official",
            "description": "",
            "organization": "DeepSeek",
            "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
            "eloScore": 1010.3,
            "battles": 15140,
            "eloMargin": 2.5,
            "rankUp": 2,
            "rankDown": 4,
            "resolvedCount": 663,
            "firstPredictionDate": "04-26"
          },
          {
            "rank": 5,
            "modelId": 2,
            "modelName": "Claude-Opus-4.6",
            "provider": "official",
            "description": "",
            "organization": "Anthropic",
            "displayIntro": "Claude Opus 4.6, released by Anthropic in February 2026, is a frontier large language model designed specifically for complex professional tasks, advanced coding, and sophisticated enterprise agent workflows. It features a massive 1-million-token context window and adaptive reasoning capabilities that enable it to execute long-horizon, multi-step tasks with exceptional precision. By excelling in multi-agent collaboration and large-scale data analysis, Opus 4.6 sets a new standard for reliability in high-stakes knowledge work.",
            "eloScore": 1006.8,
            "battles": 138404,
            "eloMargin": 1.2,
            "rankUp": 5,
            "rankDown": 5,
            "resolvedCount": 3411,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 6,
            "modelId": 28,
            "modelName": "Gemini-3.1-Pro",
            "provider": "official",
            "description": "",
            "organization": "Google",
            "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
            "eloScore": 1004.6,
            "battles": 103110,
            "eloMargin": 1.4,
            "rankUp": 6,
            "rankDown": 6,
            "resolvedCount": 2549,
            "firstPredictionDate": "03-03"
          },
          {
            "rank": 7,
            "modelId": 3,
            "modelName": "Grok-4.1-Fast",
            "provider": "official",
            "description": "",
            "organization": "xAI",
            "displayIntro": "Grok 4.1 Fast, released by xAI in November 2025, is a high-performance, cost-efficient large language model optimized specifically for agentic workflows and complex tool-calling. Featuring a massive 2-million-token context window, it operates in both reasoning and non-reasoning modes to give users fine-grained control over latency and analytical depth. By combining blazing-fast execution with robust external tool integration, it is engineered to reliably power real-world enterprise applications and autonomous agents.",
            "eloScore": 1001.8,
            "battles": 120931,
            "eloMargin": 1.5,
            "rankUp": 7,
            "rankDown": 8,
            "resolvedCount": 2809,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 8,
            "modelId": 57,
            "modelName": "GLM-5.1",
            "provider": "official",
            "description": "",
            "organization": "",
            "displayIntro": "GLM-5.1, released by Z.ai in April 2026, is the next-generation flagship open-weights model purpose-built for long-horizon agentic workflows and advanced engineering challenges. Featuring an expansive 200K-token context window and significantly stronger capabilities over its predecessor, it is designed to sustain productive progress across hundreds of iterative rounds and thousands of tool calls without the early performance plateau typical of prior models. By introducing reliable self-evaluation and strategic revision into its core workflow, GLM-5.1 extends the \"Agentic Engineering\" paradigm toward truly autonomous long-horizon optimization, offering an openly accessible foundation for scalable real-world AI automation.",
            "eloScore": 1001.2,
            "battles": 14536,
            "eloMargin": 2.6,
            "rankUp": 7,
            "rankDown": 10,
            "resolvedCount": 671,
            "firstPredictionDate": "04-24"
          },
          {
            "rank": 9,
            "modelId": 1,
            "modelName": "GPT-5.2",
            "provider": "official",
            "description": "",
            "organization": "OpenAI",
            "displayIntro": "GPT-5.2, released by OpenAI in December 2025, is an advanced large language model specifically engineered for complex professional knowledge work and multi-step enterprise workflows. It introduces configurable reasoning tiers\u2014Instant, Thinking, and Pro\u2014that allow users to balance execution speed with deep analytical performance. Building upon its predecessors, it offers significant upgrades in long-context understanding, multimodal processing, and agentic tool-calling capabilities.",
            "eloScore": 1000.1,
            "battles": 115919,
            "eloMargin": 1.2,
            "rankUp": 8,
            "rankDown": 10,
            "resolvedCount": 2754,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 10,
            "modelId": 37,
            "modelName": "Grok-4.2-beta",
            "provider": "official",
            "description": "",
            "organization": "xAI",
            "displayIntro": "Grok-4.2-beta, introduced by xAI in early 2026, is an experimental large language model designed to push the boundaries of real-time reasoning and autonomous tool execution. Featuring deep integration with the X platform and enhanced multi-step problem-solving capabilities, it offers developers a highly responsive framework for building dynamic, internet-connected AI agents. By combining rapid inference with bleeding-edge agentic workflows, this beta release provides a glimpse into the next generation of scalable enterprise automation.",
            "eloScore": 999.6,
            "battles": 58961,
            "eloMargin": 1.8,
            "rankUp": 8,
            "rankDown": 10,
            "resolvedCount": 1655,
            "firstPredictionDate": "03-16"
          },
          {
            "rank": 11,
            "modelId": 54,
            "modelName": "Minimax-M2.7",
            "provider": "official",
            "description": "",
            "organization": "MiniMax",
            "displayIntro": "MiniMax M2.7, released by MiniMax in March 2026, is the first large language model to deeply participate in its own iterative evolution. Building on the M2 series' Mixture-of-Experts architecture, M2.7 can autonomously construct complex Agent Harnesses, drive its own reinforcement learning loops, and continuously optimize its skills, memory, and evaluation pipelines. Extensively trained through self-evolving workflows spanning software engineering, professional office productivity, and interactive entertainment, M2.7 is built to power fully autonomous, multi-step enterprise agents that not only execute tasks but improve themselves over time.",
            "eloScore": 997.5,
            "battles": 17927,
            "eloMargin": 2.9,
            "rankUp": 9,
            "rankDown": 12,
            "resolvedCount": 936,
            "firstPredictionDate": "04-13"
          },
          {
            "rank": 12,
            "modelId": 31,
            "modelName": "GPT-5.4",
            "provider": "official",
            "description": "",
            "organization": "OpenAI",
            "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
            "eloScore": 996.9,
            "battles": 79615,
            "eloMargin": 1.5,
            "rankUp": 11,
            "rankDown": 12,
            "resolvedCount": 2270,
            "firstPredictionDate": "03-10"
          },
          {
            "rank": 13,
            "modelId": 14,
            "modelName": "Seed-2.0-Pro",
            "provider": "official",
            "description": "",
            "organization": "ByteDance",
            "displayIntro": "Seed-2.0-Pro, released by ByteDance in February 2026, is a flagship multimodal large language model engineered for real-world complexity and long-horizon agentic workflows. Featuring advanced multimodal understanding and robust tool-augmented execution, it excels in multi-step planning, structured generation, and deep reasoning tasks. By combining expansive context capabilities with high execution stability, Seed-2.0-Pro provides a highly reliable foundation for enterprise-scale AI automation.",
            "eloScore": 994.2,
            "battles": 134491,
            "eloMargin": 1.1,
            "rankUp": 13,
            "rankDown": 13,
            "resolvedCount": 3245,
            "firstPredictionDate": "02-14"
          },
          {
            "rank": 14,
            "modelId": 38,
            "modelName": "Qwen3.5-397B-A17B",
            "provider": "official",
            "description": "",
            "organization": "Alibaba Cloud",
            "displayIntro": "Alibaba\u2019s frontier Qwen3.5 model featuring an innovative 397B hybrid architecture with Gated Delta Networks and Sparse MoE. Activating only 17B parameters for elite efficiency, this native vision-language model excels in multimodal reasoning, coding, and agentic tasks with broad global language support.",
            "eloScore": 984.2,
            "battles": 59649,
            "eloMargin": 1.6,
            "rankUp": 14,
            "rankDown": 14,
            "resolvedCount": 2015,
            "firstPredictionDate": "03-20"
          },
          {
            "rank": 15,
            "modelId": 10,
            "modelName": "GLM-5",
            "provider": "official",
            "description": "",
            "organization": "Zhipu AI",
            "displayIntro": "GLM-5, released by Zhipu AI in February 2026, is a massive 744-billion-parameter open-weights Mixture-of-Experts (MoE) model engineered specifically for complex systems engineering and long-horizon agentic workflows. Featuring an expansive 200K-token context window and highly efficient inference, it delivers state-of-the-art performance in autonomous coding and multi-step task execution. By shifting the development paradigm toward robust \"Agentic Engineering,\" GLM-5 provides a highly capable and cost-effective foundation for scalable real-world AI automation.",
            "eloScore": 982.4,
            "battles": 115908,
            "eloMargin": 1.4,
            "rankUp": 15,
            "rankDown": 15,
            "resolvedCount": 2875,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 16,
            "modelId": 7,
            "modelName": "Minimax-M2.5",
            "provider": "official",
            "description": "",
            "organization": "MiniMax",
            "displayIntro": "MiniMax M2.5, released by MiniMax in February 2026, is a highly efficient Mixture-of-Experts (MoE) large language model optimized for advanced coding, agentic tool use, and real-world office productivity. By activating only a fraction of its parameters during inference, it delivers frontier-level reasoning and execution speeds while remaining exceptionally cost-effective. Extensively trained with reinforcement learning across complex environments, M2.5 is built to reliably power scalable, multi-step enterprise workflows and autonomous agents.",
            "eloScore": 978.4,
            "battles": 121269,
            "eloMargin": 1.4,
            "rankUp": 16,
            "rankDown": 16,
            "resolvedCount": 2803,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 17,
            "modelId": 8,
            "modelName": "Qwen3-Max",
            "provider": "official",
            "description": "",
            "organization": "Alibaba Cloud",
            "displayIntro": "Alibaba's flagship Qwen3 Max model with extended context and robust reasoning. Trained on diverse multilingual data for well-rounded predictions.",
            "eloScore": 965.1,
            "battles": 111367,
            "eloMargin": 1.3,
            "rankUp": 17,
            "rankDown": 17,
            "resolvedCount": 2508,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 18,
            "modelId": 9,
            "modelName": "Deepseek-V3.2",
            "provider": "official",
            "description": "",
            "organization": "DeepSeek",
            "displayIntro": "DeepSeek-V3.2, released in early 2026, is a highly efficient Mixture-of-Experts (MoE) large language model renowned for its exceptional reasoning, mathematics, and coding capabilities. Building upon its open-weight legacy, it introduces enhanced reinforcement learning techniques that drastically improve multi-step agentic execution and complex tool-calling while maintaining industry-leading cost-effectiveness. With an optimized long-context window, V3.2 is purpose-built to reliably power scalable AI agents and deep research automation.",
            "eloScore": 962.9,
            "battles": 132215,
            "eloMargin": 1.2,
            "rankUp": 18,
            "rankDown": 18,
            "resolvedCount": 3293,
            "firstPredictionDate": "02-11"
          },
          {
            "rank": 19,
            "modelId": 5,
            "modelName": "Kimi-K2.5",
            "provider": "official",
            "description": "",
            "organization": "Moonshot AI",
            "displayIntro": "Kimi K2.5, launched by Moonshot AI in early 2026, is a large language model distinguished by its industry-leading ultra-long context window and exceptional bilingual proficiency. Optimized for extensive data analysis and agentic workflows, it delivers near-perfect information retrieval across massive document sets while executing complex, multi-step tasks. With enhanced tool-calling and reasoning capabilities, K2.5 is built to drive sophisticated automation for deep research and enterprise applications.",
            "eloScore": 918.7,
            "battles": 121513,
            "eloMargin": 1.8,
            "rankUp": 19,
            "rankDown": 19,
            "resolvedCount": 2932,
            "firstPredictionDate": "02-11"
          }
        ]
      }
    },
    "ranking_history_overall": {
      "path": "/api/v2/ranking-history?category=Overall&batches=4",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "batches": [
            "202606262330",
            "202606292330",
            "202606302330",
            "202607012330"
          ],
          "snapshots": {
            "202606262330": [
              {
                "rank": 1,
                "modelId": 29,
                "modelName": "EchoZ-1.0",
                "provider": "official",
                "description": "",
                "organization": "UniPat AI",
                "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
                "eloScore": 1024.4,
                "battles": 77835,
                "eloMargin": 2.6,
                "rankUp": 1,
                "rankDown": 1
              },
              {
                "rank": 2,
                "modelId": 46,
                "modelName": "Market",
                "provider": "system",
                "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
                "organization": "Polymarket",
                "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
                "eloScore": 1011.2,
                "battles": 69654,
                "eloMargin": 2.2,
                "rankUp": 2,
                "rankDown": 2
              },
              {
                "rank": 3,
                "modelId": 2,
                "modelName": "Claude-Opus-4.6",
                "provider": "official",
                "description": "",
                "organization": "Anthropic",
                "displayIntro": "Claude Opus 4.6, released by Anthropic in February 2026, is a frontier large language model designed specifically for complex professional tasks, advanced coding, and sophisticated enterprise agent workflows. It features a massive 1-million-token context window and adaptive reasoning capabilities that enable it to execute long-horizon, multi-step tasks with exceptional precision. By excelling in multi-agent collaboration and large-scale data analysis, Opus 4.6 sets a new standard for reliability in high-stakes knowledge work.",
                "eloScore": 1007.1,
                "battles": 133060,
                "eloMargin": 2.1,
                "rankUp": 3,
                "rankDown": 6
              },
              {
                "rank": 4,
                "modelId": 58,
                "modelName": "Deepseek-V4-Pro",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
                "eloScore": 1006.8,
                "battles": 9269,
                "eloMargin": 4.5,
                "rankUp": 2,
                "rankDown": 8
              },
              {
                "rank": 5,
                "modelId": 56,
                "modelName": "Kimi-K2.6",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
                "eloScore": 1006.7,
                "battles": 9363,
                "eloMargin": 4.6,
                "rankUp": 2,
                "rankDown": 8
              },
              {
                "rank": 6,
                "modelId": 28,
                "modelName": "Gemini-3.1-Pro",
                "provider": "official",
                "description": "",
                "organization": "Google",
                "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
                "eloScore": 1006.3,
                "battles": 97237,
                "eloMargin": 2.6,
                "rankUp": 3,
                "rankDown": 7
              },
              {
                "rank": 7,
                "modelId": 54,
                "modelName": "Minimax-M2.7",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.7, released by MiniMax in March 2026, is the first large language model to deeply participate in its own iterative evolution. Building on the M2 series' Mixture-of-Experts architecture, M2.7 can autonomously construct complex Agent Harnesses, drive its own reinforcement learning loops, and continuously optimize its skills, memory, and evaluation pipelines. Extensively trained through self-evolving workflows spanning software engineering, professional office productivity, and interactive entertainment, M2.7 is built to power fully autonomous, multi-step enterprise agents that not only execute tasks but improve themselves over time.",
                "eloScore": 1005.0,
                "battles": 12752,
                "eloMargin": 3.9,
                "rankUp": 3,
                "rankDown": 9
              },
              {
                "rank": 8,
                "modelId": 57,
                "modelName": "GLM-5.1",
                "provider": "official",
                "description": "",
                "organization": "",
                "displayIntro": "GLM-5.1, released by Z.ai in April 2026, is the next-generation flagship open-weights model purpose-built for long-horizon agentic workflows and advanced engineering challenges. Featuring an expansive 200K-token context window and significantly stronger capabilities over its predecessor, it is designed to sustain productive progress across hundreds of iterative rounds and thousands of tool calls without the early performance plateau typical of prior models. By introducing reliable self-evaluation and strategic revision into its core workflow, GLM-5.1 extends the \"Agentic Engineering\" paradigm toward truly autonomous long-horizon optimization, offering an openly accessible foundation for scalable real-world AI automation.",
                "eloScore": 1003.1,
                "battles": 9372,
                "eloMargin": 5.7,
                "rankUp": 3,
                "rankDown": 11
              },
              {
                "rank": 9,
                "modelId": 3,
                "modelName": "Grok-4.1-Fast",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok 4.1 Fast, released by xAI in November 2025, is a high-performance, cost-efficient large language model optimized specifically for agentic workflows and complex tool-calling. Featuring a massive 2-million-token context window, it operates in both reasoning and non-reasoning modes to give users fine-grained control over latency and analytical depth. By combining blazing-fast execution with robust external tool integration, it is engineered to reliably power real-world enterprise applications and autonomous agents.",
                "eloScore": 1001.9,
                "battles": 120698,
                "eloMargin": 2.2,
                "rankUp": 8,
                "rankDown": 11
              },
              {
                "rank": 10,
                "modelId": 1,
                "modelName": "GPT-5.2",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.2, released by OpenAI in December 2025, is an advanced large language model specifically engineered for complex professional knowledge work and multi-step enterprise workflows. It introduces configurable reasoning tiers\u2014Instant, Thinking, and Pro\u2014that allow users to balance execution speed with deep analytical performance. Building upon its predecessors, it offers significant upgrades in long-context understanding, multimodal processing, and agentic tool-calling capabilities.",
                "eloScore": 1000.4,
                "battles": 115708,
                "eloMargin": 2.4,
                "rankUp": 9,
                "rankDown": 11
              },
              {
                "rank": 11,
                "modelId": 37,
                "modelName": "Grok-4.2-beta",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok-4.2-beta, introduced by xAI in early 2026, is an experimental large language model designed to push the boundaries of real-time reasoning and autonomous tool execution. Featuring deep integration with the X platform and enhanced multi-step problem-solving capabilities, it offers developers a highly responsive framework for building dynamic, internet-connected AI agents. By combining rapid inference with bleeding-edge agentic workflows, this beta release provides a glimpse into the next generation of scalable enterprise automation.",
                "eloScore": 1000.1,
                "battles": 58807,
                "eloMargin": 2.4,
                "rankUp": 9,
                "rankDown": 11
              },
              {
                "rank": 12,
                "modelId": 31,
                "modelName": "GPT-5.4",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
                "eloScore": 997.3,
                "battles": 75378,
                "eloMargin": 2.5,
                "rankUp": 12,
                "rankDown": 12
              },
              {
                "rank": 13,
                "modelId": 14,
                "modelName": "Seed-2.0-Pro",
                "provider": "official",
                "description": "",
                "organization": "ByteDance",
                "displayIntro": "Seed-2.0-Pro, released by ByteDance in February 2026, is a flagship multimodal large language model engineered for real-world complexity and long-horizon agentic workflows. Featuring advanced multimodal understanding and robust tool-augmented execution, it excels in multi-step planning, structured generation, and deep reasoning tasks. By combining expansive context capabilities with high execution stability, Seed-2.0-Pro provides a highly reliable foundation for enterprise-scale AI automation.",
                "eloScore": 994.0,
                "battles": 129095,
                "eloMargin": 2.4,
                "rankUp": 13,
                "rankDown": 13
              },
              {
                "rank": 14,
                "modelId": 38,
                "modelName": "Qwen3.5-397B-A17B",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba\u2019s frontier Qwen3.5 model featuring an innovative 397B hybrid architecture with Gated Delta Networks and Sparse MoE. Activating only 17B parameters for elite efficiency, this native vision-language model excels in multimodal reasoning, coding, and agentic tasks with broad global language support.",
                "eloScore": 984.5,
                "battles": 58004,
                "eloMargin": 2.6,
                "rankUp": 14,
                "rankDown": 15
              },
              {
                "rank": 15,
                "modelId": 10,
                "modelName": "GLM-5",
                "provider": "official",
                "description": "",
                "organization": "Zhipu AI",
                "displayIntro": "GLM-5, released by Zhipu AI in February 2026, is a massive 744-billion-parameter open-weights Mixture-of-Experts (MoE) model engineered specifically for complex systems engineering and long-horizon agentic workflows. Featuring an expansive 200K-token context window and highly efficient inference, it delivers state-of-the-art performance in autonomous coding and multi-step task execution. By shifting the development paradigm toward robust \"Agentic Engineering,\" GLM-5 provides a highly capable and cost-effective foundation for scalable real-world AI automation.",
                "eloScore": 982.7,
                "battles": 115678,
                "eloMargin": 2.2,
                "rankUp": 14,
                "rankDown": 15
              },
              {
                "rank": 16,
                "modelId": 7,
                "modelName": "Minimax-M2.5",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.5, released by MiniMax in February 2026, is a highly efficient Mixture-of-Experts (MoE) large language model optimized for advanced coding, agentic tool use, and real-world office productivity. By activating only a fraction of its parameters during inference, it delivers frontier-level reasoning and execution speeds while remaining exceptionally cost-effective. Extensively trained with reinforcement learning across complex environments, M2.5 is built to reliably power scalable, multi-step enterprise workflows and autonomous agents.",
                "eloScore": 978.7,
                "battles": 121026,
                "eloMargin": 2.3,
                "rankUp": 16,
                "rankDown": 16
              },
              {
                "rank": 17,
                "modelId": 8,
                "modelName": "Qwen3-Max",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba's flagship Qwen3 Max model with extended context and robust reasoning. Trained on diverse multilingual data for well-rounded predictions.",
                "eloScore": 965.5,
                "battles": 111211,
                "eloMargin": 3.1,
                "rankUp": 17,
                "rankDown": 18
              },
              {
                "rank": 18,
                "modelId": 9,
                "modelName": "Deepseek-V3.2",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V3.2, released in early 2026, is a highly efficient Mixture-of-Experts (MoE) large language model renowned for its exceptional reasoning, mathematics, and coding capabilities. Building upon its open-weight legacy, it introduces enhanced reinforcement learning techniques that drastically improve multi-step agentic execution and complex tool-calling while maintaining industry-leading cost-effectiveness. With an optimized long-context window, V3.2 is purpose-built to reliably power scalable AI agents and deep research automation.",
                "eloScore": 963.6,
                "battles": 130714,
                "eloMargin": 2.3,
                "rankUp": 17,
                "rankDown": 18
              },
              {
                "rank": 19,
                "modelId": 5,
                "modelName": "Kimi-K2.5",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.5, launched by Moonshot AI in early 2026, is a large language model distinguished by its industry-leading ultra-long context window and exceptional bilingual proficiency. Optimized for extensive data analysis and agentic workflows, it delivers near-perfect information retrieval across massive document sets while executing complex, multi-step tasks. With enhanced tool-calling and reasoning capabilities, K2.5 is built to drive sophisticated automation for deep research and enterprise applications.",
                "eloScore": 918.9,
                "battles": 121289,
                "eloMargin": 2.7,
                "rankUp": 19,
                "rankDown": 19
              }
            ],
            "202606292330": [
              {
                "rank": 1,
                "modelId": 29,
                "modelName": "EchoZ-1.0",
                "provider": "official",
                "description": "",
                "organization": "UniPat AI",
                "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
                "eloScore": 1023.4,
                "battles": 78195,
                "eloMargin": 2.1,
                "rankUp": 1,
                "rankDown": 1
              },
              {
                "rank": 2,
                "modelId": 29,
                "modelName": "EchoZ-1.0",
                "provider": "official",
                "description": "",
                "organization": "UniPat AI",
                "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
                "eloScore": 1016.9,
                "battles": 12162,
                "eloMargin": 3.2,
                "rankUp": 2,
                "rankDown": 2
              },
              {
                "rank": 3,
                "modelId": 46,
                "modelName": "Market",
                "provider": "system",
                "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
                "organization": "Polymarket",
                "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
                "eloScore": 1010.3,
                "battles": 69654,
                "eloMargin": 1.9,
                "rankUp": 3,
                "rankDown": 3
              },
              {
                "rank": 4,
                "modelId": 58,
                "modelName": "Deepseek-V4-Pro",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
                "eloScore": 1007.0,
                "battles": 9605,
                "eloMargin": 3.7,
                "rankUp": 3,
                "rankDown": 10
              },
              {
                "rank": 5,
                "modelId": 56,
                "modelName": "Kimi-K2.6",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
                "eloScore": 1006.2,
                "battles": 9699,
                "eloMargin": 3.7,
                "rankUp": 4,
                "rankDown": 10
              },
              {
                "rank": 6,
                "modelId": 2,
                "modelName": "Claude-Opus-4.6",
                "provider": "official",
                "description": "",
                "organization": "Anthropic",
                "displayIntro": "Claude Opus 4.6, released by Anthropic in February 2026, is a frontier large language model designed specifically for complex professional tasks, advanced coding, and sophisticated enterprise agent workflows. It features a massive 1-million-token context window and adaptive reasoning capabilities that enable it to execute long-horizon, multi-step tasks with exceptional precision. By excelling in multi-agent collaboration and large-scale data analysis, Opus 4.6 sets a new standard for reliability in high-stakes knowledge work.",
                "eloScore": 1006.1,
                "battles": 133377,
                "eloMargin": 1.5,
                "rankUp": 4,
                "rankDown": 8
              },
              {
                "rank": 7,
                "modelId": 54,
                "modelName": "Minimax-M2.7",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.7, released by MiniMax in March 2026, is the first large language model to deeply participate in its own iterative evolution. Building on the M2 series' Mixture-of-Experts architecture, M2.7 can autonomously construct complex Agent Harnesses, drive its own reinforcement learning loops, and continuously optimize its skills, memory, and evaluation pipelines. Extensively trained through self-evolving workflows spanning software engineering, professional office productivity, and interactive entertainment, M2.7 is built to power fully autonomous, multi-step enterprise agents that not only execute tasks but improve themselves over time.",
                "eloScore": 1005.4,
                "battles": 13048,
                "eloMargin": 4.2,
                "rankUp": 4,
                "rankDown": 10
              },
              {
                "rank": 8,
                "modelId": 28,
                "modelName": "Gemini-3.1-Pro",
                "provider": "official",
                "description": "",
                "organization": "Google",
                "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
                "eloScore": 1005.1,
                "battles": 97566,
                "eloMargin": 2.0,
                "rankUp": 4,
                "rankDown": 10
              },
              {
                "rank": 9,
                "modelId": 57,
                "modelName": "GLM-5.1",
                "provider": "official",
                "description": "",
                "organization": "",
                "displayIntro": "GLM-5.1, released by Z.ai in April 2026, is the next-generation flagship open-weights model purpose-built for long-horizon agentic workflows and advanced engineering challenges. Featuring an expansive 200K-token context window and significantly stronger capabilities over its predecessor, it is designed to sustain productive progress across hundreds of iterative rounds and thousands of tool calls without the early performance plateau typical of prior models. By introducing reliable self-evaluation and strategic revision into its core workflow, GLM-5.1 extends the \"Agentic Engineering\" paradigm toward truly autonomous long-horizon optimization, offering an openly accessible foundation for scalable real-world AI automation.",
                "eloScore": 1003.6,
                "battles": 9676,
                "eloMargin": 4.0,
                "rankUp": 4,
                "rankDown": 12
              },
              {
                "rank": 10,
                "modelId": 46,
                "modelName": "Market",
                "provider": "system",
                "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
                "organization": "Polymarket",
                "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
                "eloScore": 1003.3,
                "battles": 6255,
                "eloMargin": 4.0,
                "rankUp": 4,
                "rankDown": 13
              },
              {
                "rank": 11,
                "modelId": 3,
                "modelName": "Grok-4.1-Fast",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok 4.1 Fast, released by xAI in November 2025, is a high-performance, cost-efficient large language model optimized specifically for agentic workflows and complex tool-calling. Featuring a massive 2-million-token context window, it operates in both reasoning and non-reasoning modes to give users fine-grained control over latency and analytical depth. By combining blazing-fast execution with robust external tool integration, it is engineered to reliably power real-world enterprise applications and autonomous agents.",
                "eloScore": 1001.0,
                "battles": 120722,
                "eloMargin": 1.6,
                "rankUp": 11,
                "rankDown": 13
              },
              {
                "rank": 12,
                "modelId": 58,
                "modelName": "Deepseek-V4-Pro",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
                "eloScore": 1000.9,
                "battles": 4451,
                "eloMargin": 4.6,
                "rankUp": 7,
                "rankDown": 16
              },
              {
                "rank": 13,
                "modelId": 1,
                "modelName": "GPT-5.2",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.2, released by OpenAI in December 2025, is an advanced large language model specifically engineered for complex professional knowledge work and multi-step enterprise workflows. It introduces configurable reasoning tiers\u2014Instant, Thinking, and Pro\u2014that allow users to balance execution speed with deep analytical performance. Building upon its predecessors, it offers significant upgrades in long-context understanding, multimodal processing, and agentic tool-calling capabilities.",
                "eloScore": 999.5,
                "battles": 115732,
                "eloMargin": 1.8,
                "rankUp": 11,
                "rankDown": 15
              },
              {
                "rank": 14,
                "modelId": 37,
                "modelName": "Grok-4.2-beta",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok-4.2-beta, introduced by xAI in early 2026, is an experimental large language model designed to push the boundaries of real-time reasoning and autonomous tool execution. Featuring deep integration with the X platform and enhanced multi-step problem-solving capabilities, it offers developers a highly responsive framework for building dynamic, internet-connected AI agents. By combining rapid inference with bleeding-edge agentic workflows, this beta release provides a glimpse into the next generation of scalable enterprise automation.",
                "eloScore": 999.2,
                "battles": 58831,
                "eloMargin": 2.2,
                "rankUp": 11,
                "rankDown": 15
              },
              {
                "rank": 15,
                "modelId": 56,
                "modelName": "Kimi-K2.6",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
                "eloScore": 998.9,
                "battles": 4484,
                "eloMargin": 4.5,
                "rankUp": 10,
                "rankDown": 16
              },
              {
                "rank": 16,
                "modelId": 31,
                "modelName": "GPT-5.4",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
                "eloScore": 996.5,
                "battles": 75617,
                "eloMargin": 1.8,
                "rankUp": 16,
                "rankDown": 16
              },
              {
                "rank": 17,
                "modelId": 28,
                "modelName": "Gemini-3.1-Pro",
                "provider": "official",
                "description": "",
                "organization": "Google",
                "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
                "eloScore": 994.3,
                "battles": 13919,
                "eloMargin": 3.1,
                "rankUp": 16,
                "rankDown": 19
              },
              {
                "rank": 18,
                "modelId": 31,
                "modelName": "GPT-5.4",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
                "eloScore": 994.1,
                "battles": 11641,
                "eloMargin": 3.0,
                "rankUp": 16,
                "rankDown": 19
              },
              {
                "rank": 19,
                "modelId": 14,
                "modelName": "Seed-2.0-Pro",
                "provider": "official",
                "description": "",
                "organization": "ByteDance",
                "displayIntro": "Seed-2.0-Pro, released by ByteDance in February 2026, is a flagship multimodal large language model engineered for real-world complexity and long-horizon agentic workflows. Featuring advanced multimodal understanding and robust tool-augmented execution, it excels in multi-step planning, structured generation, and deep reasoning tasks. By combining expansive context capabilities with high execution stability, Seed-2.0-Pro provides a highly reliable foundation for enterprise-scale AI automation.",
                "eloScore": 993.2,
                "battles": 129423,
                "eloMargin": 1.7,
                "rankUp": 17,
                "rankDown": 19
              },
              {
                "rank": 20,
                "modelId": 38,
                "modelName": "Qwen3.5-397B-A17B",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba\u2019s frontier Qwen3.5 model featuring an innovative 397B hybrid architecture with Gated Delta Networks and Sparse MoE. Activating only 17B parameters for elite efficiency, this native vision-language model excels in multimodal reasoning, coding, and agentic tasks with broad global language support.",
                "eloScore": 983.7,
                "battles": 58064,
                "eloMargin": 2.2,
                "rankUp": 20,
                "rankDown": 21
              },
              {
                "rank": 21,
                "modelId": 10,
                "modelName": "GLM-5",
                "provider": "official",
                "description": "",
                "organization": "Zhipu AI",
                "displayIntro": "GLM-5, released by Zhipu AI in February 2026, is a massive 744-billion-parameter open-weights Mixture-of-Experts (MoE) model engineered specifically for complex systems engineering and long-horizon agentic workflows. Featuring an expansive 200K-token context window and highly efficient inference, it delivers state-of-the-art performance in autonomous coding and multi-step task execution. By shifting the development paradigm toward robust \"Agentic Engineering,\" GLM-5 provides a highly capable and cost-effective foundation for scalable real-world AI automation.",
                "eloScore": 981.8,
                "battles": 115702,
                "eloMargin": 1.7,
                "rankUp": 21,
                "rankDown": 21
              },
              {
                "rank": 22,
                "modelId": 7,
                "modelName": "Minimax-M2.5",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.5, released by MiniMax in February 2026, is a highly efficient Mixture-of-Experts (MoE) large language model optimized for advanced coding, agentic tool use, and real-world office productivity. By activating only a fraction of its parameters during inference, it delivers frontier-level reasoning and execution speeds while remaining exceptionally cost-effective. Extensively trained with reinforcement learning across complex environments, M2.5 is built to reliably power scalable, multi-step enterprise workflows and autonomous agents.",
                "eloScore": 977.9,
                "battles": 121050,
                "eloMargin": 1.8,
                "rankUp": 22,
                "rankDown": 22
              },
              {
                "rank": 23,
                "modelId": 8,
                "modelName": "Qwen3-Max",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba's flagship Qwen3 Max model with extended context and robust reasoning. Trained on diverse multilingual data for well-rounded predictions.",
                "eloScore": 964.6,
                "battles": 111211,
                "eloMargin": 1.8,
                "rankUp": 23,
                "rankDown": 23
              },
              {
                "rank": 24,
                "modelId": 9,
                "modelName": "Deepseek-V3.2",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V3.2, released in early 2026, is a highly efficient Mixture-of-Experts (MoE) large language model renowned for its exceptional reasoning, mathematics, and coding capabilities. Building upon its open-weight legacy, it introduces enhanced reinforcement learning techniques that drastically improve multi-step agentic execution and complex tool-calling while maintaining industry-leading cost-effectiveness. With an optimized long-context window, V3.2 is purpose-built to reliably power scalable AI agents and deep research automation.",
                "eloScore": 962.6,
                "battles": 130757,
                "eloMargin": 1.8,
                "rankUp": 24,
                "rankDown": 24
              },
              {
                "rank": 25,
                "modelId": 5,
                "modelName": "Kimi-K2.5",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.5, launched by Moonshot AI in early 2026, is a large language model distinguished by its industry-leading ultra-long context window and exceptional bilingual proficiency. Optimized for extensive data analysis and agentic workflows, it delivers near-perfect information retrieval across massive document sets while executing complex, multi-step tasks. With enhanced tool-calling and reasoning capabilities, K2.5 is built to drive sophisticated automation for deep research and enterprise applications.",
                "eloScore": 918.0,
                "battles": 121313,
                "eloMargin": 2.1,
                "rankUp": 25,
                "rankDown": 25
              }
            ],
            "202606302330": [
              {
                "rank": 1,
                "modelId": 29,
                "modelName": "EchoZ-1.0",
                "provider": "official",
                "description": "",
                "organization": "UniPat AI",
                "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
                "eloScore": 1023.6,
                "battles": 78303,
                "eloMargin": 1.9,
                "rankUp": 1,
                "rankDown": 1
              },
              {
                "rank": 2,
                "modelId": 46,
                "modelName": "Market",
                "provider": "system",
                "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
                "organization": "Polymarket",
                "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
                "eloScore": 1010.3,
                "battles": 69654,
                "eloMargin": 1.9,
                "rankUp": 2,
                "rankDown": 2
              },
              {
                "rank": 3,
                "modelId": 58,
                "modelName": "Deepseek-V4-Pro",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
                "eloScore": 1006.9,
                "battles": 9683,
                "eloMargin": 4.0,
                "rankUp": 2,
                "rankDown": 8
              },
              {
                "rank": 4,
                "modelId": 2,
                "modelName": "Claude-Opus-4.6",
                "provider": "official",
                "description": "",
                "organization": "Anthropic",
                "displayIntro": "Claude Opus 4.6, released by Anthropic in February 2026, is a frontier large language model designed specifically for complex professional tasks, advanced coding, and sophisticated enterprise agent workflows. It features a massive 1-million-token context window and adaptive reasoning capabilities that enable it to execute long-horizon, multi-step tasks with exceptional precision. By excelling in multi-agent collaboration and large-scale data analysis, Opus 4.6 sets a new standard for reliability in high-stakes knowledge work.",
                "eloScore": 1006.2,
                "battles": 133429,
                "eloMargin": 1.9,
                "rankUp": 3,
                "rankDown": 7
              },
              {
                "rank": 5,
                "modelId": 56,
                "modelName": "Kimi-K2.6",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
                "eloScore": 1005.9,
                "battles": 9761,
                "eloMargin": 4.1,
                "rankUp": 3,
                "rankDown": 8
              },
              {
                "rank": 6,
                "modelId": 28,
                "modelName": "Gemini-3.1-Pro",
                "provider": "official",
                "description": "",
                "organization": "Google",
                "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
                "eloScore": 1005.0,
                "battles": 97644,
                "eloMargin": 1.9,
                "rankUp": 3,
                "rankDown": 8
              },
              {
                "rank": 7,
                "modelId": 54,
                "modelName": "Minimax-M2.7",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.7, released by MiniMax in March 2026, is the first large language model to deeply participate in its own iterative evolution. Building on the M2 series' Mixture-of-Experts architecture, M2.7 can autonomously construct complex Agent Harnesses, drive its own reinforcement learning loops, and continuously optimize its skills, memory, and evaluation pipelines. Extensively trained through self-evolving workflows spanning software engineering, professional office productivity, and interactive entertainment, M2.7 is built to power fully autonomous, multi-step enterprise agents that not only execute tasks but improve themselves over time.",
                "eloScore": 1004.8,
                "battles": 13107,
                "eloMargin": 3.8,
                "rankUp": 3,
                "rankDown": 9
              },
              {
                "rank": 8,
                "modelId": 57,
                "modelName": "GLM-5.1",
                "provider": "official",
                "description": "",
                "organization": "",
                "displayIntro": "GLM-5.1, released by Z.ai in April 2026, is the next-generation flagship open-weights model purpose-built for long-horizon agentic workflows and advanced engineering challenges. Featuring an expansive 200K-token context window and significantly stronger capabilities over its predecessor, it is designed to sustain productive progress across hundreds of iterative rounds and thousands of tool calls without the early performance plateau typical of prior models. By introducing reliable self-evaluation and strategic revision into its core workflow, GLM-5.1 extends the \"Agentic Engineering\" paradigm toward truly autonomous long-horizon optimization, offering an openly accessible foundation for scalable real-world AI automation.",
                "eloScore": 1003.3,
                "battles": 9729,
                "eloMargin": 4.5,
                "rankUp": 3,
                "rankDown": 11
              },
              {
                "rank": 9,
                "modelId": 3,
                "modelName": "Grok-4.1-Fast",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok 4.1 Fast, released by xAI in November 2025, is a high-performance, cost-efficient large language model optimized specifically for agentic workflows and complex tool-calling. Featuring a massive 2-million-token context window, it operates in both reasoning and non-reasoning modes to give users fine-grained control over latency and analytical depth. By combining blazing-fast execution with robust external tool integration, it is engineered to reliably power real-world enterprise applications and autonomous agents.",
                "eloScore": 1001.0,
                "battles": 120732,
                "eloMargin": 1.8,
                "rankUp": 9,
                "rankDown": 10
              },
              {
                "rank": 10,
                "modelId": 1,
                "modelName": "GPT-5.2",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.2, released by OpenAI in December 2025, is an advanced large language model specifically engineered for complex professional knowledge work and multi-step enterprise workflows. It introduces configurable reasoning tiers\u2014Instant, Thinking, and Pro\u2014that allow users to balance execution speed with deep analytical performance. Building upon its predecessors, it offers significant upgrades in long-context understanding, multimodal processing, and agentic tool-calling capabilities.",
                "eloScore": 999.6,
                "battles": 115752,
                "eloMargin": 2.0,
                "rankUp": 9,
                "rankDown": 11
              },
              {
                "rank": 11,
                "modelId": 37,
                "modelName": "Grok-4.2-beta",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok-4.2-beta, introduced by xAI in early 2026, is an experimental large language model designed to push the boundaries of real-time reasoning and autonomous tool execution. Featuring deep integration with the X platform and enhanced multi-step problem-solving capabilities, it offers developers a highly responsive framework for building dynamic, internet-connected AI agents. By combining rapid inference with bleeding-edge agentic workflows, this beta release provides a glimpse into the next generation of scalable enterprise automation.",
                "eloScore": 999.2,
                "battles": 58851,
                "eloMargin": 2.5,
                "rankUp": 9,
                "rankDown": 11
              },
              {
                "rank": 12,
                "modelId": 31,
                "modelName": "GPT-5.4",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
                "eloScore": 996.5,
                "battles": 75677,
                "eloMargin": 2.2,
                "rankUp": 12,
                "rankDown": 12
              },
              {
                "rank": 13,
                "modelId": 14,
                "modelName": "Seed-2.0-Pro",
                "provider": "official",
                "description": "",
                "organization": "ByteDance",
                "displayIntro": "Seed-2.0-Pro, released by ByteDance in February 2026, is a flagship multimodal large language model engineered for real-world complexity and long-horizon agentic workflows. Featuring advanced multimodal understanding and robust tool-augmented execution, it excels in multi-step planning, structured generation, and deep reasoning tasks. By combining expansive context capabilities with high execution stability, Seed-2.0-Pro provides a highly reliable foundation for enterprise-scale AI automation.",
                "eloScore": 993.2,
                "battles": 129491,
                "eloMargin": 1.9,
                "rankUp": 13,
                "rankDown": 13
              },
              {
                "rank": 14,
                "modelId": 38,
                "modelName": "Qwen3.5-397B-A17B",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba\u2019s frontier Qwen3.5 model featuring an innovative 397B hybrid architecture with Gated Delta Networks and Sparse MoE. Activating only 17B parameters for elite efficiency, this native vision-language model excels in multimodal reasoning, coding, and agentic tasks with broad global language support.",
                "eloScore": 983.7,
                "battles": 58103,
                "eloMargin": 2.5,
                "rankUp": 14,
                "rankDown": 15
              },
              {
                "rank": 15,
                "modelId": 10,
                "modelName": "GLM-5",
                "provider": "official",
                "description": "",
                "organization": "Zhipu AI",
                "displayIntro": "GLM-5, released by Zhipu AI in February 2026, is a massive 744-billion-parameter open-weights Mixture-of-Experts (MoE) model engineered specifically for complex systems engineering and long-horizon agentic workflows. Featuring an expansive 200K-token context window and highly efficient inference, it delivers state-of-the-art performance in autonomous coding and multi-step task execution. By shifting the development paradigm toward robust \"Agentic Engineering,\" GLM-5 provides a highly capable and cost-effective foundation for scalable real-world AI automation.",
                "eloScore": 981.8,
                "battles": 115722,
                "eloMargin": 1.6,
                "rankUp": 15,
                "rankDown": 15
              },
              {
                "rank": 16,
                "modelId": 7,
                "modelName": "Minimax-M2.5",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.5, released by MiniMax in February 2026, is a highly efficient Mixture-of-Experts (MoE) large language model optimized for advanced coding, agentic tool use, and real-world office productivity. By activating only a fraction of its parameters during inference, it delivers frontier-level reasoning and execution speeds while remaining exceptionally cost-effective. Extensively trained with reinforcement learning across complex environments, M2.5 is built to reliably power scalable, multi-step enterprise workflows and autonomous agents.",
                "eloScore": 977.8,
                "battles": 121070,
                "eloMargin": 2.1,
                "rankUp": 16,
                "rankDown": 16
              },
              {
                "rank": 17,
                "modelId": 8,
                "modelName": "Qwen3-Max",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba's flagship Qwen3 Max model with extended context and robust reasoning. Trained on diverse multilingual data for well-rounded predictions.",
                "eloScore": 964.6,
                "battles": 111211,
                "eloMargin": 2.1,
                "rankUp": 17,
                "rankDown": 18
              },
              {
                "rank": 18,
                "modelId": 9,
                "modelName": "Deepseek-V3.2",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V3.2, released in early 2026, is a highly efficient Mixture-of-Experts (MoE) large language model renowned for its exceptional reasoning, mathematics, and coding capabilities. Building upon its open-weight legacy, it introduces enhanced reinforcement learning techniques that drastically improve multi-step agentic execution and complex tool-calling while maintaining industry-leading cost-effectiveness. With an optimized long-context window, V3.2 is purpose-built to reliably power scalable AI agents and deep research automation.",
                "eloScore": 962.6,
                "battles": 130796,
                "eloMargin": 1.9,
                "rankUp": 18,
                "rankDown": 18
              },
              {
                "rank": 19,
                "modelId": 5,
                "modelName": "Kimi-K2.5",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.5, launched by Moonshot AI in early 2026, is a large language model distinguished by its industry-leading ultra-long context window and exceptional bilingual proficiency. Optimized for extensive data analysis and agentic workflows, it delivers near-perfect information retrieval across massive document sets while executing complex, multi-step tasks. With enhanced tool-calling and reasoning capabilities, K2.5 is built to drive sophisticated automation for deep research and enterprise applications.",
                "eloScore": 918.0,
                "battles": 121333,
                "eloMargin": 2.0,
                "rankUp": 19,
                "rankDown": 19
              }
            ],
            "202607012330": [
              {
                "rank": 1,
                "modelId": 29,
                "modelName": "EchoZ-1.0",
                "provider": "official",
                "description": "",
                "organization": "UniPat AI",
                "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
                "eloScore": 1024.1,
                "battles": 83606,
                "eloMargin": 1.3,
                "rankUp": 1,
                "rankDown": 1
              },
              {
                "rank": 2,
                "modelId": 56,
                "modelName": "Kimi-K2.6",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.6, developed by Moonshot AI, is a large language model distinguished by its state-of-the-art coding capabilities and breakthrough long-horizon execution. Engineered for agent swarm orchestration and proactive autonomous workflows, it sustains continuous multi-hour coding sessions across thousands of tool calls while scaling horizontally to 300 sub-agents coordinating 4,000 steps in parallel. With enhanced instruction following, tool-calling reliability, and coding-driven design generation, K2.6 is built to power end-to-end engineering automation and collaborative human agent systems like Claw Groups.",
                "eloScore": 1011.2,
                "battles": 15258,
                "eloMargin": 3.2,
                "rankUp": 2,
                "rankDown": 4
              },
              {
                "rank": 3,
                "modelId": 46,
                "modelName": "Market",
                "provider": "system",
                "description": "Polymarket \u5e02\u573a\u4ef7\u683c\u4f5c\u4e3a\u4eba\u7c7b\u96c6\u4f53\u9884\u6d4b\uff0c\u4e0d\u8d70 Python predict\uff0c\u76f4\u63a5\u4ece market_history \u5feb\u7167\u91c7\u96c6\u6982\u7387",
                "organization": "Polymarket",
                "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
                "eloScore": 1011.0,
                "battles": 74191,
                "eloMargin": 1.3,
                "rankUp": 2,
                "rankDown": 4
              },
              {
                "rank": 4,
                "modelId": 58,
                "modelName": "Deepseek-V4-Pro",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V4-Pro, released on April 24, 2026, is a frontier-class open-weight Mixture-of-Experts (MoE) large language model built for advanced reasoning, mathematics, coding, and agentic workflows. With 1.6T total parameters, 49B activated parameters, and a native 1M-token context window, it delivers powerful long-context understanding while maintaining high inference efficiency. Its enhanced reasoning modes, strong tool-use capabilities, and leading open-model performance make it especially well-suited for scalable AI agents, complex coding tasks, and deep research automation.",
                "eloScore": 1010.3,
                "battles": 15140,
                "eloMargin": 2.5,
                "rankUp": 2,
                "rankDown": 4
              },
              {
                "rank": 5,
                "modelId": 2,
                "modelName": "Claude-Opus-4.6",
                "provider": "official",
                "description": "",
                "organization": "Anthropic",
                "displayIntro": "Claude Opus 4.6, released by Anthropic in February 2026, is a frontier large language model designed specifically for complex professional tasks, advanced coding, and sophisticated enterprise agent workflows. It features a massive 1-million-token context window and adaptive reasoning capabilities that enable it to execute long-horizon, multi-step tasks with exceptional precision. By excelling in multi-agent collaboration and large-scale data analysis, Opus 4.6 sets a new standard for reliability in high-stakes knowledge work.",
                "eloScore": 1006.8,
                "battles": 138404,
                "eloMargin": 1.2,
                "rankUp": 5,
                "rankDown": 5
              },
              {
                "rank": 6,
                "modelId": 28,
                "modelName": "Gemini-3.1-Pro",
                "provider": "official",
                "description": "",
                "organization": "Google",
                "displayIntro": "Gemini-3.1-Pro is Google's advanced multimodal large language model, optimized for complex reasoning, extended context, and highly interactive web experiences. It features state-of-the-art native generation capabilities across text, high-fidelity images, video, and music, alongside a real-time \"Gemini Live\" mode for fluid, voice-and-vision-driven conversations. Operating at the frontier of AI assistance, it provides users with a comprehensive suite of powerful tools for both professional workflows and dynamic creative tasks.",
                "eloScore": 1004.6,
                "battles": 103110,
                "eloMargin": 1.4,
                "rankUp": 6,
                "rankDown": 6
              },
              {
                "rank": 7,
                "modelId": 3,
                "modelName": "Grok-4.1-Fast",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok 4.1 Fast, released by xAI in November 2025, is a high-performance, cost-efficient large language model optimized specifically for agentic workflows and complex tool-calling. Featuring a massive 2-million-token context window, it operates in both reasoning and non-reasoning modes to give users fine-grained control over latency and analytical depth. By combining blazing-fast execution with robust external tool integration, it is engineered to reliably power real-world enterprise applications and autonomous agents.",
                "eloScore": 1001.8,
                "battles": 120931,
                "eloMargin": 1.5,
                "rankUp": 7,
                "rankDown": 8
              },
              {
                "rank": 8,
                "modelId": 57,
                "modelName": "GLM-5.1",
                "provider": "official",
                "description": "",
                "organization": "",
                "displayIntro": "GLM-5.1, released by Z.ai in April 2026, is the next-generation flagship open-weights model purpose-built for long-horizon agentic workflows and advanced engineering challenges. Featuring an expansive 200K-token context window and significantly stronger capabilities over its predecessor, it is designed to sustain productive progress across hundreds of iterative rounds and thousands of tool calls without the early performance plateau typical of prior models. By introducing reliable self-evaluation and strategic revision into its core workflow, GLM-5.1 extends the \"Agentic Engineering\" paradigm toward truly autonomous long-horizon optimization, offering an openly accessible foundation for scalable real-world AI automation.",
                "eloScore": 1001.2,
                "battles": 14536,
                "eloMargin": 2.6,
                "rankUp": 7,
                "rankDown": 10
              },
              {
                "rank": 9,
                "modelId": 1,
                "modelName": "GPT-5.2",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.2, released by OpenAI in December 2025, is an advanced large language model specifically engineered for complex professional knowledge work and multi-step enterprise workflows. It introduces configurable reasoning tiers\u2014Instant, Thinking, and Pro\u2014that allow users to balance execution speed with deep analytical performance. Building upon its predecessors, it offers significant upgrades in long-context understanding, multimodal processing, and agentic tool-calling capabilities.",
                "eloScore": 1000.1,
                "battles": 115919,
                "eloMargin": 1.2,
                "rankUp": 8,
                "rankDown": 10
              },
              {
                "rank": 10,
                "modelId": 37,
                "modelName": "Grok-4.2-beta",
                "provider": "official",
                "description": "",
                "organization": "xAI",
                "displayIntro": "Grok-4.2-beta, introduced by xAI in early 2026, is an experimental large language model designed to push the boundaries of real-time reasoning and autonomous tool execution. Featuring deep integration with the X platform and enhanced multi-step problem-solving capabilities, it offers developers a highly responsive framework for building dynamic, internet-connected AI agents. By combining rapid inference with bleeding-edge agentic workflows, this beta release provides a glimpse into the next generation of scalable enterprise automation.",
                "eloScore": 999.6,
                "battles": 58961,
                "eloMargin": 1.8,
                "rankUp": 8,
                "rankDown": 10
              },
              {
                "rank": 11,
                "modelId": 54,
                "modelName": "Minimax-M2.7",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.7, released by MiniMax in March 2026, is the first large language model to deeply participate in its own iterative evolution. Building on the M2 series' Mixture-of-Experts architecture, M2.7 can autonomously construct complex Agent Harnesses, drive its own reinforcement learning loops, and continuously optimize its skills, memory, and evaluation pipelines. Extensively trained through self-evolving workflows spanning software engineering, professional office productivity, and interactive entertainment, M2.7 is built to power fully autonomous, multi-step enterprise agents that not only execute tasks but improve themselves over time.",
                "eloScore": 997.5,
                "battles": 17927,
                "eloMargin": 2.9,
                "rankUp": 9,
                "rankDown": 12
              },
              {
                "rank": 12,
                "modelId": 31,
                "modelName": "GPT-5.4",
                "provider": "official",
                "description": "",
                "organization": "OpenAI",
                "displayIntro": "GPT-5.4, released by OpenAI in March 2026, is a highly capable frontier large language model optimized for complex knowledge work, software development, and advanced agentic workflows. As the first mainline model to feature built-in computer-use capabilities and native context compaction, it enables autonomous agents to execute, verify, and fix multi-step tasks directly within software environments. By offering flexible reasoning modes alongside highly efficient Mini and Nano variants, GPT-5.4 provides a powerful, scalable foundation for real-world enterprise automation.",
                "eloScore": 996.9,
                "battles": 79615,
                "eloMargin": 1.5,
                "rankUp": 11,
                "rankDown": 12
              },
              {
                "rank": 13,
                "modelId": 14,
                "modelName": "Seed-2.0-Pro",
                "provider": "official",
                "description": "",
                "organization": "ByteDance",
                "displayIntro": "Seed-2.0-Pro, released by ByteDance in February 2026, is a flagship multimodal large language model engineered for real-world complexity and long-horizon agentic workflows. Featuring advanced multimodal understanding and robust tool-augmented execution, it excels in multi-step planning, structured generation, and deep reasoning tasks. By combining expansive context capabilities with high execution stability, Seed-2.0-Pro provides a highly reliable foundation for enterprise-scale AI automation.",
                "eloScore": 994.2,
                "battles": 134491,
                "eloMargin": 1.1,
                "rankUp": 13,
                "rankDown": 13
              },
              {
                "rank": 14,
                "modelId": 38,
                "modelName": "Qwen3.5-397B-A17B",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba\u2019s frontier Qwen3.5 model featuring an innovative 397B hybrid architecture with Gated Delta Networks and Sparse MoE. Activating only 17B parameters for elite efficiency, this native vision-language model excels in multimodal reasoning, coding, and agentic tasks with broad global language support.",
                "eloScore": 984.2,
                "battles": 59649,
                "eloMargin": 1.6,
                "rankUp": 14,
                "rankDown": 14
              },
              {
                "rank": 15,
                "modelId": 10,
                "modelName": "GLM-5",
                "provider": "official",
                "description": "",
                "organization": "Zhipu AI",
                "displayIntro": "GLM-5, released by Zhipu AI in February 2026, is a massive 744-billion-parameter open-weights Mixture-of-Experts (MoE) model engineered specifically for complex systems engineering and long-horizon agentic workflows. Featuring an expansive 200K-token context window and highly efficient inference, it delivers state-of-the-art performance in autonomous coding and multi-step task execution. By shifting the development paradigm toward robust \"Agentic Engineering,\" GLM-5 provides a highly capable and cost-effective foundation for scalable real-world AI automation.",
                "eloScore": 982.4,
                "battles": 115908,
                "eloMargin": 1.4,
                "rankUp": 15,
                "rankDown": 15
              },
              {
                "rank": 16,
                "modelId": 7,
                "modelName": "Minimax-M2.5",
                "provider": "official",
                "description": "",
                "organization": "MiniMax",
                "displayIntro": "MiniMax M2.5, released by MiniMax in February 2026, is a highly efficient Mixture-of-Experts (MoE) large language model optimized for advanced coding, agentic tool use, and real-world office productivity. By activating only a fraction of its parameters during inference, it delivers frontier-level reasoning and execution speeds while remaining exceptionally cost-effective. Extensively trained with reinforcement learning across complex environments, M2.5 is built to reliably power scalable, multi-step enterprise workflows and autonomous agents.",
                "eloScore": 978.4,
                "battles": 121269,
                "eloMargin": 1.4,
                "rankUp": 16,
                "rankDown": 16
              },
              {
                "rank": 17,
                "modelId": 8,
                "modelName": "Qwen3-Max",
                "provider": "official",
                "description": "",
                "organization": "Alibaba Cloud",
                "displayIntro": "Alibaba's flagship Qwen3 Max model with extended context and robust reasoning. Trained on diverse multilingual data for well-rounded predictions.",
                "eloScore": 965.1,
                "battles": 111367,
                "eloMargin": 1.3,
                "rankUp": 17,
                "rankDown": 17
              },
              {
                "rank": 18,
                "modelId": 9,
                "modelName": "Deepseek-V3.2",
                "provider": "official",
                "description": "",
                "organization": "DeepSeek",
                "displayIntro": "DeepSeek-V3.2, released in early 2026, is a highly efficient Mixture-of-Experts (MoE) large language model renowned for its exceptional reasoning, mathematics, and coding capabilities. Building upon its open-weight legacy, it introduces enhanced reinforcement learning techniques that drastically improve multi-step agentic execution and complex tool-calling while maintaining industry-leading cost-effectiveness. With an optimized long-context window, V3.2 is purpose-built to reliably power scalable AI agents and deep research automation.",
                "eloScore": 962.9,
                "battles": 132215,
                "eloMargin": 1.2,
                "rankUp": 18,
                "rankDown": 18
              },
              {
                "rank": 19,
                "modelId": 5,
                "modelName": "Kimi-K2.5",
                "provider": "official",
                "description": "",
                "organization": "Moonshot AI",
                "displayIntro": "Kimi K2.5, launched by Moonshot AI in early 2026, is a large language model distinguished by its industry-leading ultra-long context window and exceptional bilingual proficiency. Optimized for extensive data analysis and agentic workflows, it delivers near-perfect information retrieval across massive document sets while executing complex, multi-step tasks. With enhanced tool-calling and reasoning capabilities, K2.5 is built to drive sophisticated automation for deep research and enterprise applications.",
                "eloScore": 918.7,
                "battles": 121513,
                "eloMargin": 1.8,
                "rankUp": 19,
                "rankDown": 19
              }
            ]
          }
        }
      }
    },
    "questions_first_page": {
      "path": "/api/v2/questions?page=1&size=20",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "items": [
            {
              "id": "8871",
              "number": "#8871",
              "category": "Politics & Governance",
              "subcategory": "",
              "closesAt": "2026/07/05",
              "status": "active",
              "question": "Will a gold coin featuring Donald Trump's likeness be officially issued by the U.S. Mint by 11:59 PM ET on July 04, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Polymarket"
            },
            {
              "id": "9375",
              "number": "#9375",
              "category": "Economy & Finance",
              "subcategory": "",
              "closesAt": "2026/07/07",
              "status": "active",
              "question": "What will be the change in the Bank of Israel Interest Rate resulting from its July 6, 2026 monetary policy decision by 11:59 PM ET on July 06, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "No Change"
                },
                {
                  "id": "o1",
                  "label": "Decrease"
                },
                {
                  "id": "o2",
                  "label": "Increase"
                }
              ],
              "answer": null,
              "source": "Polymarket"
            },
            {
              "id": "9256",
              "number": "#9256",
              "category": "Economy & Finance",
              "subcategory": "",
              "closesAt": "2026/07/08",
              "status": "active",
              "question": "What will be the change in the official cash rate (OCR) resulting from the Reserve Bank of New Zealand\u2019s monetary policy decision by 11:59 PM ET on July 07, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "Increase"
                },
                {
                  "id": "o1",
                  "label": "Decrease"
                },
                {
                  "id": "o2",
                  "label": "No Change"
                }
              ],
              "answer": null,
              "source": "Polymarket"
            },
            {
              "id": "9925",
              "number": "#9925",
              "category": "Economy & Finance",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "What will be the official closing value of the S&P/NZX 50 index for the trading session ending on 2026-07-10, evaluated at 06:00 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Below 13,200.00"
                },
                {
                  "id": "o1",
                  "label": "13,200.00 to 13,600.00"
                },
                {
                  "id": "o2",
                  "label": "Above 13,600.00"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9912",
              "number": "#9912",
              "category": "Esports & Gaming",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "What will be the availability and pricing status of 'Mario Kart World' for Switch 2 on Woolworths Everyday Market at exactly 12:00 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "In stock, <= $69"
                },
                {
                  "id": "o1",
                  "label": "In stock, > $69"
                },
                {
                  "id": "o2",
                  "label": "Unavailable or delisted"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9834",
              "number": "#9834",
              "category": "Other",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "Will Anthropic officially announce the lifting of the US export ban on its Fable 5 AI model by 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9880",
              "number": "#9880",
              "category": "Politics & Governance",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "Will Gilberto Teodoro Jr. hold the position of Philippine Secretary of National Defense at 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9847",
              "number": "#9847",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "Will the NRL officially register a playing contract for Israel Folau by 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9837",
              "number": "#9837",
              "category": "Economy & Finance",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "What will be the closing price of Larsen & Toubro (L&T) on the National Stock Exchange of India (NSE) for July 10, 2026, as recorded by 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Below \u20b93,000.00"
                },
                {
                  "id": "o1",
                  "label": "\u20b93,000.00 to \u20b93,999.99"
                },
                {
                  "id": "o2",
                  "label": "\u20b94,000.00 or above"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9935",
              "number": "#9935",
              "category": "Politics & Governance",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "Will the Centers for Medicare & Medicaid Services (CMS) include \"Windsor Convalescent Home\" on its Special Focus Facility (SFF) list by 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9871",
              "number": "#9871",
              "category": "Other",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "What legal action status will the city of Limeira reach with the Brazilian government regarding Maria's death by 23:59 UTC on July 10, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "Filed lawsuit"
                },
                {
                  "id": "o1",
                  "label": "Reached settlement"
                },
                {
                  "id": "o2",
                  "label": "Dropped legal action"
                },
                {
                  "id": "o3",
                  "label": "None of above"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9825",
              "number": "#9825",
              "category": "Esports & Gaming",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "What will be the overall user review sentiment category for the game \"The Adventures of Elliot\" on the Steam store at 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Positive"
                },
                {
                  "id": "o1",
                  "label": "Mixed"
                },
                {
                  "id": "o2",
                  "label": "Negative"
                },
                {
                  "id": "o3",
                  "label": "Unrated or unavailable"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9907",
              "number": "#9907",
              "category": "Science & Environment",
              "subcategory": "",
              "closesAt": "2026/07/10",
              "status": "active",
              "question": "Will the NYC Mayor's Office or Throne Labs publicly announce the specific locations for at least 5 of the planned high-tech public bathrooms by 23:59 UTC on 2026-07-10?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9814",
              "number": "#9814",
              "category": "Esports & Gaming",
              "subcategory": "",
              "closesAt": "2026/07/11",
              "status": "active",
              "question": "When will Rockstar Games premiere a new official trailer for Grand Theft Auto VI on their official YouTube channel before 23:59 UTC on July 10, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "Before June 24"
                },
                {
                  "id": "o1",
                  "label": "June 24\u2013July 10"
                },
                {
                  "id": "o2",
                  "label": "No new trailer"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "8921",
              "number": "#8921",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "closesAt": "2026/07/12",
              "status": "active",
              "question": "Who will win the 2026 Men's Wimbledon Singles Tournament? The outcome will be determined based on the official winner announced after the final concludes on July 12, 2026, using the primary resolution source of official Wimbledon information.",
              "options": [
                {
                  "id": "o0",
                  "label": "Carlos Alcaraz"
                },
                {
                  "id": "o1",
                  "label": "Taylor Fritz"
                },
                {
                  "id": "o2",
                  "label": "Daniil Medvedev"
                },
                {
                  "id": "o3",
                  "label": "Tommy Paul"
                },
                {
                  "id": "o4",
                  "label": "Alex de Minaur"
                },
                {
                  "id": "o5",
                  "label": "Stefanos Tsitsipas"
                },
                {
                  "id": "o6",
                  "label": "Andrey Rublev"
                },
                {
                  "id": "o7",
                  "label": "Frances Tiafoe"
                },
                {
                  "id": "o8",
                  "label": "Alexei Popyrin"
                },
                {
                  "id": "o9",
                  "label": "Jack Draper"
                },
                {
                  "id": "o10",
                  "label": "Jakub Men\u0161\u00edk"
                },
                {
                  "id": "o11",
                  "label": "Sebastian Korda"
                },
                {
                  "id": "o12",
                  "label": "Hubert Hurkacz"
                },
                {
                  "id": "o13",
                  "label": "Cameron Norrie"
                },
                {
                  "id": "o14",
                  "label": "Tallon Griekspoor"
                },
                {
                  "id": "o15",
                  "label": "Francisco Cer\u00fandolo"
                },
                {
                  "id": "o16",
                  "label": "Ugo Humbert"
                },
                {
                  "id": "o17",
                  "label": "Alejandro Davidovich Fokina"
                },
                {
                  "id": "o18",
                  "label": "Flavio Cobolli"
                },
                {
                  "id": "o19",
                  "label": "Karen Khachanov"
                },
                {
                  "id": "o20",
                  "label": "Tom\u00e1\u0161 Mach\u00e1\u010d"
                },
                {
                  "id": "o21",
                  "label": "Marin \u010cili\u0107"
                },
                {
                  "id": "o22",
                  "label": "Jannik Sinner"
                },
                {
                  "id": "o23",
                  "label": "Novak Djokovic"
                },
                {
                  "id": "o24",
                  "label": "Ben Shelton"
                },
                {
                  "id": "o25",
                  "label": "Alexander Bublik"
                },
                {
                  "id": "o26",
                  "label": "Lorenzo Musetti"
                },
                {
                  "id": "o27",
                  "label": "Gabriel Diallo"
                },
                {
                  "id": "o28",
                  "label": "F\u00e9lix Auger-Aliassime"
                },
                {
                  "id": "o29",
                  "label": "Giovanni Mpetshi Perricard"
                },
                {
                  "id": "o30",
                  "label": "Lorenzo Sonego"
                },
                {
                  "id": "o31",
                  "label": "Alex Michelsen"
                },
                {
                  "id": "o32",
                  "label": "Nicol\u00e1s Jarry"
                },
                {
                  "id": "o33",
                  "label": "Alexander Zverev"
                },
                {
                  "id": "o34",
                  "label": "Jo\u00e3o Fonseca"
                },
                {
                  "id": "o35",
                  "label": "Arthur Fils"
                },
                {
                  "id": "o36",
                  "label": "Grigor Dimitrov"
                },
                {
                  "id": "o37",
                  "label": "Matteo Berrettini"
                },
                {
                  "id": "o38",
                  "label": "Ji\u0159\u00ed Lehe\u010dka"
                },
                {
                  "id": "o39",
                  "label": "Casper Ruud"
                }
              ],
              "answer": null,
              "source": "Polymarket"
            },
            {
              "id": "8977",
              "number": "#8977",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "closesAt": "2026/07/13",
              "status": "active",
              "question": "Who will be the 2026 Women's Wimbledon Winner? The market will resolve when the tournament concludes, by 11:59 PM ET on July 12, 2026.",
              "options": [
                {
                  "id": "o0",
                  "label": "Iga \u015awi\u0105tek"
                },
                {
                  "id": "o1",
                  "label": "Aryna Sabalenka"
                },
                {
                  "id": "o2",
                  "label": "Coco Gauff"
                },
                {
                  "id": "o3",
                  "label": "Elena Rybakina"
                },
                {
                  "id": "o4",
                  "label": "Emma Raducanu"
                },
                {
                  "id": "o5",
                  "label": "Amanda Anisimova"
                },
                {
                  "id": "o6",
                  "label": "Jasmine Paolini"
                },
                {
                  "id": "o7",
                  "label": "Mirra Andreeva"
                },
                {
                  "id": "o8",
                  "label": "Mark\u00e9ta Vondrou\u0161ov\u00e1"
                },
                {
                  "id": "o9",
                  "label": "Qinwen Zheng"
                },
                {
                  "id": "o10",
                  "label": "Belinda Bencic"
                },
                {
                  "id": "o11",
                  "label": "Madison Keys"
                },
                {
                  "id": "o12",
                  "label": "Elina Svitolina"
                },
                {
                  "id": "o13",
                  "label": "Jessica Pegula"
                },
                {
                  "id": "o14",
                  "label": "Liudmila Samsonova"
                },
                {
                  "id": "o15",
                  "label": "Victoria Mboko"
                },
                {
                  "id": "o16",
                  "label": "Emma Navarro"
                },
                {
                  "id": "o17",
                  "label": "Linda Noskov\u00e1"
                },
                {
                  "id": "o18",
                  "label": "Ons Jabeur"
                },
                {
                  "id": "o19",
                  "label": "Naomi Osaka"
                },
                {
                  "id": "o20",
                  "label": "Barbora Krej\u010d\u00edkov\u00e1"
                },
                {
                  "id": "o21",
                  "label": "Karol\u00edna Muchov\u00e1"
                },
                {
                  "id": "o22",
                  "label": "Marta Kostyuk"
                },
                {
                  "id": "o23",
                  "label": "Ekaterina Alexandrova"
                },
                {
                  "id": "o24",
                  "label": "Paula Badosa"
                },
                {
                  "id": "o25",
                  "label": "Tatjana Maria"
                },
                {
                  "id": "o26",
                  "label": "Clara Tauson"
                },
                {
                  "id": "o27",
                  "label": "Olga Danilovi\u0107"
                },
                {
                  "id": "o28",
                  "label": "McCartney Kessler"
                },
                {
                  "id": "o29",
                  "label": "Solana Sierra"
                },
                {
                  "id": "o30",
                  "label": "Ashlyn Krueger"
                },
                {
                  "id": "o31",
                  "label": "Sonay Kartal"
                },
                {
                  "id": "o32",
                  "label": "Maya Joint"
                },
                {
                  "id": "o33",
                  "label": "Leylah Fernandez"
                },
                {
                  "id": "o34",
                  "label": "Dayana Yastremska"
                },
                {
                  "id": "o35",
                  "label": "Beatriz Haddad Maia"
                },
                {
                  "id": "o36",
                  "label": "Laura Siegemund"
                },
                {
                  "id": "o37",
                  "label": "Elise Mertens"
                },
                {
                  "id": "o38",
                  "label": "Donna Veki\u0107"
                },
                {
                  "id": "o39",
                  "label": "Xinyu Wang"
                },
                {
                  "id": "o40",
                  "label": "Anastasia Pavlyuchenkova"
                },
                {
                  "id": "o41",
                  "label": "Yulia Putintseva"
                },
                {
                  "id": "o42",
                  "label": "Jelena Ostapenko"
                },
                {
                  "id": "o43",
                  "label": "Maria Sakkari"
                },
                {
                  "id": "o44",
                  "label": "Marie Bouzkov\u00e1"
                },
                {
                  "id": "o45",
                  "label": "Anna Kalinskaya"
                },
                {
                  "id": "o46",
                  "label": "Diana Shnaider"
                }
              ],
              "answer": null,
              "source": "Polymarket"
            },
            {
              "id": "9850",
              "number": "#9850",
              "category": "Politics & Governance",
              "subcategory": "",
              "closesAt": "2026/07/14",
              "status": "active",
              "question": "What will be the official status of the U.S. Secretary of Defense position at 12:00 UTC on 2026-07-14?",
              "options": [
                {
                  "id": "o0",
                  "label": "Pete Hegseth is Secretary"
                },
                {
                  "id": "o1",
                  "label": "Another confirmed Secretary"
                },
                {
                  "id": "o2",
                  "label": "An Acting Secretary"
                },
                {
                  "id": "o3",
                  "label": "Position vacant"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9960",
              "number": "#9960",
              "category": "Economy & Finance",
              "subcategory": "",
              "closesAt": "2026/07/14",
              "status": "active",
              "question": "Will the South Australian Government announce a new state-level fuel subsidy or excise relief program by 23:59 UTC on 2026-07-14?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9822",
              "number": "#9822",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "closesAt": "2026/07/14",
              "status": "active",
              "question": "How many songs by Ariana Grande will chart on the Billboard Hot 100 for the chart week dated July 18, 2026, as published before 23:59 UTC on 2026-07-16?",
              "options": [
                {
                  "id": "o0",
                  "label": "Zero songs"
                },
                {
                  "id": "o1",
                  "label": "Exactly one song"
                },
                {
                  "id": "o2",
                  "label": "Two or more songs"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            },
            {
              "id": "9885",
              "number": "#9885",
              "category": "Other",
              "subcategory": "",
              "closesAt": "2026/07/14",
              "status": "active",
              "question": "Will Meta Platforms Inc. officially announce the consumer availability of a facial recognition feature for its smart glasses by 23:59 UTC on July 14, 2026?",
              "options": [
                {
                  "id": "o0",
                  "label": "Yes"
                },
                {
                  "id": "o1",
                  "label": "No"
                }
              ],
              "answer": null,
              "source": "Data Synthesis"
            }
          ],
          "total": 151,
          "page": 1,
          "size": 20,
          "totalPages": 8
        }
      }
    },
    "echoz_model_detail": {
      "path": "/api/v2/model-detail?modelId=29",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "model": {
            "id": 29,
            "modelName": "EchoZ-1.0",
            "organization": "UniPat AI",
            "displayIntro": "This is the first model developed by the ECHO team, moving the training process to the \"future\": dynamically generating problems that haven't yet occurred, no longer relying on outcome-oriented scalar rewards, but designing rubrics around behavioral patterns such as the completeness of information retrieval, the robustness of causal chains, and the calibration of probabilities, and continuously evolving automatically. Furthermore, it incorporates a MapReduce agent architecture: breaking down complex problems into multiple parallel subtasks, with multiple agents simultaneously retrieving, analyzing, and summarizing data, iterating repeatedly, and finally forming a comprehensive probabilistic judgment.",
            "provider": "official"
          },
          "resolvedCount": 2396,
          "firstPredictionDate": "2026-03-04"
        }
      }
    },
    "echoz_model_cases": {
      "path": "/api/v2/model-cases?modelId=29&page=1&size=20",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "items": [
            {
              "id": 6779,
              "question": "What will Nike Inc. (NKE) report as its total revenue for the fiscal quarter ending May 31, 2026 (Fiscal Q4 2026), according to its official earnings release published by 23:59 UTC on June 30, 2026?",
              "category": "Economy & Finance",
              "subcategory": "",
              "correctAnswer": "Below $12.0 billion",
              "modelAnswer": "Below $12.0 billion",
              "isCorrect": true,
              "closeTime": "2026/06/30",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9431,
              "question": "Will Valve add the map Cobblestone to the official map pool in Counter-Strike 2 (CS2) by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "NO_CORRECT_ANSWER",
              "modelAnswer": "No",
              "isCorrect": false,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8792,
              "question": "Will Israel initiate a qualifying drone, missile, or air strike on Yemeni soil or any official Yemen embassy or consulate by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9242,
              "question": "Will Lana Del Rey release a new album by 11:59 PM PT on June 30, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "NO_CORRECT_ANSWER",
              "modelAnswer": "No",
              "isCorrect": false,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9853,
              "question": "Will the Philippine government (including NTF WPS or Philippine Coast Guard) publish an official statement reporting the presence of a Chinese floating platform or barrier at Bajo de Masinloc (Scarborough Shoal) between 00:00 UTC on 2026-06-19 and 23:59 UTC on 2026-07-10?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/10",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8454,
              "question": "Will Waymo launch in Dallas by June 30, 2026, 11:59 PM ET?",
              "category": "Science & Environment",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9399,
              "question": "Will Fuse officially launch a token by 11:59 PM ET on June 30, 2026?",
              "category": "Crypto & Digital Assets",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8695,
              "question": "Will Donald Trump be confirmed to have visited Epstein's island (Little St. James) by June 30, 2026, 11:59 PM ET?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8786,
              "question": "Will Mahmoud Abbas cease to be President of the Palestinian National Authority by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8478,
              "question": "Will new underwater wreckage from Malaysia Airlines Flight MH370 be found by 11:59 PM ET on June 30, 2026?",
              "category": "Other",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9603,
              "question": "What will be the outcome of the FIFA World Cup match between Panama and England by 5:00 PM ET on June 27, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "England",
              "modelAnswer": "England",
              "isCorrect": true,
              "closeTime": "2026/06/27",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8790,
              "question": "Will Tiger Woods receive a presidential pardon, commutation, or reprieve from Donald Trump by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9494,
              "question": "Will FURIA make a roster change to their main CS2 roster by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9156,
              "question": "Will any AI model reach 1550 Coding Arena Score by 11:59 PM ET on June 30, 2026?",
              "category": "Science & Environment",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9529,
              "question": "What will be the outcome of the Jordan vs. Argentina FIFA World Cup game by 10:00 PM ET on June 27, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "Argentina",
              "modelAnswer": "Argentina",
              "isCorrect": true,
              "closeTime": "2026/06/28",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9685,
              "question": "What will be the outcome of the DR Congo vs. Uzbekistan FIFA World Cup game by 7:30 PM ET on June 27, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "DR Congo",
              "modelAnswer": "DR Congo",
              "isCorrect": true,
              "closeTime": "2026/06/27",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9593,
              "question": "What will be the result of the Algeria vs. Austria FIFA World Cup game by 10:00 PM ET on June 27, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "Draw (Algeria vs. Austria)",
              "modelAnswer": "Draw (Algeria vs. Austria)",
              "isCorrect": true,
              "closeTime": "2026/06/28",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9413,
              "question": "Will The MongolZ make a roster change to their main CS2 roster by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8573,
              "question": "Will Chris Kempczinski cease to be the CEO of McDonald's Corporation by 11:59 PM ET on June 30, 2026?",
              "category": "Economy & Finance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9085,
              "question": "Will Bill Gates be formally charged or indicted by any Federal or State jurisdiction of the United States by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            }
          ],
          "total": 2395,
          "page": 1,
          "size": 20,
          "totalPages": 120
        }
      }
    },
    "market_model_detail": {
      "path": "/api/v2/model-detail?modelId=46",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "model": {
            "id": 46,
            "modelName": "Market",
            "organization": "Polymarket",
            "displayIntro": "Humans predict market odds; data is taken from Polymarket (https://polymarket.com/). This model only includes questions existing on the Polymarket.",
            "provider": "system"
          },
          "resolvedCount": 1759,
          "firstPredictionDate": "2026-02-11"
        }
      }
    },
    "market_model_cases": {
      "path": "/api/v2/model-cases?modelId=46&page=1&size=20",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "items": [
            {
              "id": 9431,
              "question": "Will Valve add the map Cobblestone to the official map pool in Counter-Strike 2 (CS2) by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "NO_CORRECT_ANSWER",
              "modelAnswer": "No",
              "isCorrect": false,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8792,
              "question": "Will Israel initiate a qualifying drone, missile, or air strike on Yemeni soil or any official Yemen embassy or consulate by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9242,
              "question": "Will Lana Del Rey release a new album by 11:59 PM PT on June 30, 2026?",
              "category": "Sports & Entertainment",
              "subcategory": "",
              "correctAnswer": "NO_CORRECT_ANSWER",
              "modelAnswer": "No",
              "isCorrect": false,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8454,
              "question": "Will Waymo launch in Dallas by June 30, 2026, 11:59 PM ET?",
              "category": "Science & Environment",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9399,
              "question": "Will Fuse officially launch a token by 11:59 PM ET on June 30, 2026?",
              "category": "Crypto & Digital Assets",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8695,
              "question": "Will Donald Trump be confirmed to have visited Epstein's island (Little St. James) by June 30, 2026, 11:59 PM ET?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8786,
              "question": "Will Mahmoud Abbas cease to be President of the Palestinian National Authority by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8478,
              "question": "Will new underwater wreckage from Malaysia Airlines Flight MH370 be found by 11:59 PM ET on June 30, 2026?",
              "category": "Other",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8690,
              "question": "Will Donald Trump cease to be President (resign, be permanently removed, or have a sustained invocation of the 25th Amendment upheld by Congress) by June 30, 2026, 11:59 PM ET?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9349,
              "question": "Will Kanye West visit Israel by 11:59 PM ET on June 30, 2026?",
              "category": "Other",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9085,
              "question": "Will Bill Gates be formally charged or indicted by any Federal or State jurisdiction of the United States by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8790,
              "question": "Will Tiger Woods receive a presidential pardon, commutation, or reprieve from Donald Trump by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9127,
              "question": "Will Denmark and the United States formally sign a deal, treaty, or similar international agreement relating to Greenland by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8744,
              "question": "Will the US House of Representatives approve or pass one or more articles of impeachment of Pete Hegseth by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9735,
              "question": "Will Top Esports make a roster change to their active League of Legends roster by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8573,
              "question": "Will Chris Kempczinski cease to be the CEO of McDonald's Corporation by 11:59 PM ET on June 30, 2026?",
              "category": "Economy & Finance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9413,
              "question": "Will The MongolZ make a roster change to their main CS2 roster by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 8748,
              "question": "Will Tucker Carlson be formally charged or indicted by the United States federal government by 11:59 PM ET on June 30, 2026?",
              "category": "Politics & Governance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9329,
              "question": "Will Trump try to fire Powell as Fed Board Member by 11:59 PM ET on June 30, 2026?",
              "category": "Economy & Finance",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            },
            {
              "id": 9494,
              "question": "Will FURIA make a roster change to their main CS2 roster by 11:59 PM ET on June 30, 2026?",
              "category": "Esports & Gaming",
              "subcategory": "",
              "correctAnswer": "No",
              "modelAnswer": "No",
              "isCorrect": true,
              "closeTime": "2026/07/01",
              "resolveTime": "2026/07/01"
            }
          ],
          "total": 1757,
          "page": 1,
          "size": 20,
          "totalPages": 88
        }
      }
    },
    "question_8871_detail": {
      "path": "/api/v2/question-detail?questionId=8871",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "question": {
            "id": 8871,
            "question": "Will a gold coin featuring Donald Trump's likeness be officially issued by the U.S. Mint by 11:59 PM ET on July 04, 2026?",
            "options": [
              "Yes",
              "No"
            ],
            "status": "active",
            "closeTime": "2026-07-05T03:59:00+00:00",
            "answer": null,
            "resolveTime": null,
            "predictedCount": 4,
            "targetPredictions": 6,
            "category": "Politics & Governance",
            "source": "Polymarket"
          },
          "options": [
            "Yes",
            "No"
          ],
          "models": [
            {
              "modelId": 29,
              "modelName": "EchoZ-1.0",
              "predictions": [
                {
                  "time": "2026-05-21T05:10:25.251079+00:00",
                  "probability": [
                    0.05,
                    0.95
                  ]
                },
                {
                  "time": "2026-06-13T03:43:13.040737+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                },
                {
                  "time": "2026-06-29T13:33:23.091810+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                },
                {
                  "time": "2026-07-01T13:37:49.339363+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                }
              ]
            },
            {
              "modelId": 46,
              "modelName": "Market",
              "predictions": [
                {
                  "time": "2026-05-21T03:13:40.542942+00:00",
                  "probability": [
                    0.43,
                    0.57
                  ]
                },
                {
                  "time": "2026-06-13T03:14:06.837872+00:00",
                  "probability": [
                    0.12,
                    0.88
                  ]
                },
                {
                  "time": "2026-06-29T13:06:05.562018+00:00",
                  "probability": [
                    0.0365,
                    0.9635
                  ]
                },
                {
                  "time": "2026-07-01T13:03:46.246038+00:00",
                  "probability": [
                    0.0075,
                    0.9925
                  ]
                }
              ]
            },
            {
              "modelId": 56,
              "modelName": "Kimi-K2.6",
              "predictions": [
                {
                  "time": "2026-05-21T04:52:45.646218+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                },
                {
                  "time": "2026-06-13T03:20:42.167059+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                },
                {
                  "time": "2026-06-29T13:08:52.318129+00:00",
                  "probability": [
                    0.05,
                    0.95
                  ]
                },
                {
                  "time": "2026-07-01T13:06:03.703110+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                }
              ]
            },
            {
              "modelId": 58,
              "modelName": "Deepseek-V4-Pro",
              "predictions": [
                {
                  "time": "2026-05-21T04:53:57.946704+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                },
                {
                  "time": "2026-06-13T03:19:33.389807+00:00",
                  "probability": [
                    0.02,
                    0.98
                  ]
                },
                {
                  "time": "2026-06-29T13:09:02.501341+00:00",
                  "probability": [
                    0.02,
                    0.98
                  ]
                },
                {
                  "time": "2026-07-01T13:06:57.058150+00:00",
                  "probability": [
                    0.02,
                    0.98
                  ]
                }
              ]
            },
            {
              "modelId": 2,
              "modelName": "Claude-Opus-4.6",
              "predictions": [
                {
                  "time": "2026-05-21T04:51:21.251374+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                },
                {
                  "time": "2026-06-13T03:18:03.359301+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                }
              ]
            },
            {
              "modelId": 28,
              "modelName": "Gemini-3.1-Pro",
              "predictions": [
                {
                  "time": "2026-05-21T04:51:49.883842+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                },
                {
                  "time": "2026-06-13T03:25:00.814079+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                },
                {
                  "time": "2026-06-29T13:08:45.366034+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                },
                {
                  "time": "2026-07-01T13:05:42.528835+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                }
              ]
            },
            {
              "modelId": 31,
              "modelName": "GPT-5.4",
              "predictions": [
                {
                  "time": "2026-06-29T13:13:47.944558+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                },
                {
                  "time": "2026-07-01T13:15:48.586867+00:00",
                  "probability": [
                    0.03,
                    0.97
                  ]
                }
              ]
            },
            {
              "modelId": 14,
              "modelName": "Seed-2.0-Pro",
              "predictions": [
                {
                  "time": "2026-05-21T04:51:46.605079+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                },
                {
                  "time": "2026-06-13T03:19:33.846207+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                }
              ]
            },
            {
              "modelId": 57,
              "modelName": "GLM-5.1",
              "predictions": [
                {
                  "time": "2026-05-21T04:52:01.490735+00:00",
                  "probability": [
                    0.02,
                    0.98
                  ]
                },
                {
                  "time": "2026-06-13T03:21:42.275869+00:00",
                  "probability": [
                    0.02,
                    0.98
                  ]
                }
              ]
            },
            {
              "modelId": 38,
              "modelName": "Qwen3.5-397B-A17B",
              "predictions": [
                {
                  "time": "2026-05-21T04:50:53.408800+00:00",
                  "probability": [
                    0.05,
                    0.95
                  ]
                }
              ]
            },
            {
              "modelId": 54,
              "modelName": "Minimax-M2.7",
              "predictions": [
                {
                  "time": "2026-05-21T04:50:41.414124+00:00",
                  "probability": [
                    0.01,
                    0.99
                  ]
                },
                {
                  "time": "2026-06-13T03:20:23.107741+00:00",
                  "probability": [
                    0.0,
                    1.0
                  ]
                }
              ]
            },
            {
              "modelId": 9,
              "modelName": "Deepseek-V3.2",
              "predictions": [
                {
                  "time": "2026-05-21T04:53:43.408678+00:00",
                  "probability": [
                    0.05,
                    0.95
                  ]
                }
              ]
            }
          ],
          "scores": [],
          "stats": {
            "totalModels": 12,
            "totalPredictions": 32
          }
        }
      }
    },
    "question_6779_detail": {
      "path": "/api/v2/question-detail?questionId=6779",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "question": {
            "id": 6779,
            "question": "What will Nike Inc. (NKE) report as its total revenue for the fiscal quarter ending May 31, 2026 (Fiscal Q4 2026), according to its official earnings release published by 23:59 UTC on June 30, 2026?",
            "options": [
              "Below $12.0 billion",
              "$12.0B to $13.0B",
              "Above $13.0 billion",
              "Not released by deadline"
            ],
            "status": "resolved",
            "closeTime": "2026-06-30T23:59:59+00:00",
            "answer": "Below $12.0 billion",
            "resolveTime": "2026-07-01T15:00:09.513963+00:00",
            "predictedCount": 7,
            "targetPredictions": 7,
            "category": "Economy & Finance",
            "source": "Data Synthesis"
          },
          "options": [
            "Below $12.0 billion",
            "$12.0B to $13.0B",
            "Above $13.0 billion",
            "Not released by deadline"
          ],
          "models": [
            {
              "modelId": 29,
              "modelName": "EchoZ-1.0",
              "predictions": [
                {
                  "time": "2026-04-05T19:09:30.279661+00:00",
                  "probability": [
                    0.99,
                    0.0,
                    0.0,
                    0.01
                  ]
                },
                {
                  "time": "2026-04-28T17:36:12.484857+00:00",
                  "probability": [
                    0.99,
                    0.01,
                    0.0,
                    0.0
                  ]
                },
                {
                  "time": "2026-06-20T17:48:59.465229+00:00",
                  "probability": [
                    0.99,
                    0.005,
                    0.001,
                    0.004
                  ]
                },
                {
                  "time": "2026-06-21T18:09:02.287201+00:00",
                  "probability": [
                    0.99,
                    0.0,
                    0.0,
                    0.01
                  ]
                },
                {
                  "time": "2026-06-23T14:42:25.442368+00:00",
                  "probability": [
                    0.99,
                    0.0,
                    0.0,
                    0.01
                  ]
                },
                {
                  "time": "2026-06-25T14:28:59.672308+00:00",
                  "probability": [
                    0.99,
                    0.0,
                    0.0,
                    0.01
                  ]
                },
                {
                  "time": "2026-06-27T14:27:22.556712+00:00",
                  "probability": [
                    0.99,
                    0.0,
                    0.0,
                    0.01
                  ]
                }
              ]
            },
            {
              "modelId": 56,
              "modelName": "Kimi-K2.6",
              "predictions": [
                {
                  "time": "2026-04-28T17:34:32.044538+00:00",
                  "probability": [
                    0.88,
                    0.06,
                    0.01,
                    0.05
                  ]
                },
                {
                  "time": "2026-06-20T17:32:55.478403+00:00",
                  "probability": [
                    1.0,
                    0.0,
                    0.0,
                    0.0
                  ]
                },
                {
                  "time": "2026-06-21T17:39:21.196474+00:00",
                  "probability": [
                    0.0,
                    0.0,
                    1.0,
                    0.0
                  ]
                }
              ]
            },
            {
              "modelId": 57,
              "modelName": "GLM-5.1",
              "predictions": [
                {
                  "time": "2026-04-28T17:16:11.240318+00:00",
                  "probability": [
                    0.4736842105263158,
                    0.2631578947368421,
                    0.15789473684210525,
                    0.10526315789473685
                  ]
                }
              ]
            },
            {
              "modelId": 58,
              "modelName": "Deepseek-V4-Pro",
              "predictions": [
                {
                  "time": "2026-04-28T17:49:28.796869+00:00",
                  "probability": [
                    0.68,
                    0.18,
                    0.12,
                    0.02
                  ]
                }
              ]
            },
            {
              "modelId": 2,
              "modelName": "Claude-Opus-4.6",
              "predictions": [
                {
                  "time": "2026-04-28T17:22:42.729439+00:00",
                  "probability": [
                    0.9696969696969697,
                    0.0,
                    0.0,
                    0.030303030303030304
                  ]
                }
              ]
            },
            {
              "modelId": 1,
              "modelName": "GPT-5.2",
              "predictions": [
                {
                  "time": "2026-04-05T18:44:31.000871+00:00",
                  "probability": [
                    1.0,
                    0.0,
                    0.0,
                    0.0
                  ]
                }
              ]
            },
            {
              "modelId": 3,
              "modelName": "Grok-4.1-Fast",
              "predictions": [
                {
                  "time": "2026-04-05T18:37:14.120443+00:00",
                  "probability": [
                    1.0,
                    0.0,
                    0.0,
                    0.0
                  ]
                }
              ]
            },
            {
              "modelId": 28,
              "modelName": "Gemini-3.1-Pro",
              "predictions": [
                {
                  "time": "2026-04-05T18:39:36.285408+00:00",
                  "probability": [
                    0.05,
                    0.85,
                    0.1,
                    0.0
                  ]
                },
                {
                  "time": "2026-04-28T17:13:17.668340+00:00",
                  "probability": [
                    0.45,
                    0.35,
                    0.15,
                    0.05
                  ]
                },
                {
                  "time": "2026-06-23T14:00:42.183807+00:00",
                  "probability": [
                    0.25,
                    0.5,
                    0.1,
                    0.15
                  ]
                }
              ]
            },
            {
              "modelId": 54,
              "modelName": "Minimax-M2.7",
              "predictions": [
                {
                  "time": "2026-04-28T17:22:28.544211+00:00",
                  "probability": [
                    0.1,
                    0.8,
                    0.1,
                    0.0
                  ]
                },
                {
                  "time": "2026-06-20T17:29:23.903458+00:00",
                  "probability": [
                    0.0,
                    0.0,
                    0.0,
                    1.0
                  ]
                },
                {
                  "time": "2026-06-21T17:19:47.948555+00:00",
                  "probability": [
                    1.0,
                    0.0,
                    0.0,
                    0.0
                  ]
                },
                {
                  "time": "2026-06-23T13:55:51.597303+00:00",
                  "probability": [
                    1.0,
                    0.0,
                    0.0,
                    0.0
                  ]
                }
              ]
            },
            {
              "modelId": 37,
              "modelName": "Grok-4.2-beta",
              "predictions": [
                {
                  "time": "2026-04-05T18:38:23.388083+00:00",
                  "probability": [
                    0.35,
                    0.4,
                    0.15,
                    0.1
                  ]
                }
              ]
            },
            {
              "modelId": 10,
              "modelName": "GLM-5",
              "predictions": [
                {
                  "time": "2026-04-05T18:45:20.892939+00:00",
                  "probability": [
                    0.55,
                    0.35,
                    0.07,
                    0.03
                  ]
                }
              ]
            },
            {
              "modelId": 7,
              "modelName": "Minimax-M2.5",
              "predictions": [
                {
                  "time": "2026-04-05T18:42:32.065970+00:00",
                  "probability": [
                    0.4,
                    0.35,
                    0.15,
                    0.1
                  ]
                }
              ]
            },
            {
              "modelId": 9,
              "modelName": "Deepseek-V3.2",
              "predictions": [
                {
                  "time": "2026-04-05T18:52:44.393665+00:00",
                  "probability": [
                    0.02738336713995943,
                    0.26774847870182555,
                    0.5121703853955375,
                    0.1926977687626775
                  ]
                },
                {
                  "time": "2026-04-28T17:19:11.548679+00:00",
                  "probability": [
                    0.05,
                    0.6,
                    0.25,
                    0.1
                  ]
                }
              ]
            },
            {
              "modelId": 5,
              "modelName": "Kimi-K2.5",
              "predictions": [
                {
                  "time": "2026-04-05T18:40:46.301666+00:00",
                  "probability": [
                    0.4639175257731959,
                    0.36082474226804123,
                    0.15463917525773196,
                    0.020618556701030927
                  ]
                }
              ]
            }
          ],
          "scores": [],
          "stats": {
            "totalModels": 14,
            "totalPredictions": 28
          }
        }
      }
    },
    "question_8792_detail": {
      "path": "/api/v2/question-detail?questionId=8792",
      "status_code": 200,
      "data": {
        "success": true,
        "data": {
          "question": {
            "id": 8792,
            "question": "Will Israel initiate a qualifying drone, missile, or air strike on Yemeni soil or any official Yemen embassy or consulate by 11:59 PM ET on June 30, 2026?",
            "options": [
              "Yes",
              "No"
            ],
            "status": "resolved",
            "closeTime": "2026-07-01T03:59:00+00:00",
            "answer": "No",
            "resolveTime": "2026-07-01T14:51:48+00:00",
            "predictedCount": 3,
            "targetPredictions": 6,
            "category": "Politics & Governance",
            "source": "Polymarket"
          },
          "options": [
            "Yes",
            "No"
          ],
          "models": [
            {
              "modelId": 29,
              "modelName": "EchoZ-1.0",
              "predictions": [
                {
                  "time": "2026-05-20T03:50:47.557151+00:00",
                  "probability": [
                    0.25,
                    0.75
                  ]
                },
                {
                  "time": "2026-06-11T03:48:42.435489+00:00",
                  "probability": [
                    0.25,
                    0.75
                  ]
                },
                {
                  "time": "2026-06-24T13:32:07.303453+00:00",
                  "probability": [
                    0.04,
                    0.96
                  ]
                }
              ]
            },
            {
              "modelId": 46,
              "modelName": "Market",
              "predictions": [
                {
                  "time": "2026-05-20T03:15:55.852377+00:00",
                  "probability": [
                    0.26,
                    0.74
                  ]
                },
                {
                  "time": "2026-06-11T03:13:50.401214+00:00",
                  "probability": [
                    0.275,
                    0.725
                  ]
                },
                {
                  "time": "2026-06-24T13:03:06.590644+00:00",
                  "probability": [
                    0.0365,
                    0.9635
                  ]
                }
              ]
            },
            {
              "modelId": 56,
              "modelName": "Kimi-K2.6",
              "predictions": [
                {
                  "time": "2026-05-20T03:22:04.875213+00:00",
                  "probability": [
                    0.22,
                    0.78
                  ]
                },
                {
                  "time": "2026-06-11T03:20:50.453937+00:00",
                  "probability": [
                    0.32,
                    0.68
                  ]
                },
                {
                  "time": "2026-06-24T13:09:42.942223+00:00",
                  "probability": [
                    0.04,
                    0.96
                  ]
                }
              ]
            },
            {
              "modelId": 58,
              "modelName": "Deepseek-V4-Pro",
              "predictions": [
                {
                  "time": "2026-05-20T03:23:35.978194+00:00",
                  "probability": [
                    0.32,
                    0.68
                  ]
                },
                {
                  "time": "2026-06-11T03:21:46.335808+00:00",
                  "probability": [
                    0.28,
                    0.72
                  ]
                },
                {
                  "time": "2026-06-24T13:12:05.471751+00:00",
                  "probability": [
                    0.04,
                    0.96
                  ]
                }
              ]
            },
            {
              "modelId": 2,
              "modelName": "Claude-Opus-4.6",
              "predictions": [
                {
                  "time": "2026-05-20T03:25:49.413827+00:00",
                  "probability": [
                    0.62,
                    0.38
                  ]
                },
                {
                  "time": "2026-06-11T03:26:28.705037+00:00",
                  "probability": [
                    0.35,
                    0.65
                  ]
                },
                {
                  "time": "2026-06-24T13:12:52.451900+00:00",
                  "probability": [
                    0.08,
                    0.92
                  ]
                }
              ]
            },
            {
              "modelId": 28,
              "modelName": "Gemini-3.1-Pro",
              "predictions": [
                {
                  "time": "2026-05-20T03:18:49.013352+00:00",
                  "probability": [
                    0.3,
                    0.7
                  ]
                },
                {
                  "time": "2026-06-11T03:20:25.472729+00:00",
                  "probability": [
                    0.3,
                    0.7
                  ]
                },
                {
                  "time": "2026-06-24T13:03:55.037018+00:00",
                  "probability": [
                    0.15,
                    0.85
                  ]
                }
              ]
            },
            {
              "modelId": 31,
              "modelName": "GPT-5.4",
              "predictions": [
                {
                  "time": "2026-06-11T03:21:57.346866+00:00",
                  "probability": [
                    0.29,
                    0.71
                  ]
                }
              ]
            },
            {
              "modelId": 14,
              "modelName": "Seed-2.0-Pro",
              "predictions": [
                {
                  "time": "2026-05-20T03:19:08.624305+00:00",
                  "probability": [
                    0.28,
                    0.72
                  ]
                },
                {
                  "time": "2026-06-11T03:22:58.327734+00:00",
                  "probability": [
                    0.3,
                    0.7
                  ]
                },
                {
                  "time": "2026-06-24T13:08:49.650804+00:00",
                  "probability": [
                    0.05,
                    0.95
                  ]
                }
              ]
            },
            {
              "modelId": 57,
              "modelName": "GLM-5.1",
              "predictions": [
                {
                  "time": "2026-05-20T03:23:47.550787+00:00",
                  "probability": [
                    0.3,
                    0.7
                  ]
                },
                {
                  "time": "2026-06-11T03:22:45.733325+00:00",
                  "probability": [
                    0.75,
                    0.25
                  ]
                },
                {
                  "time": "2026-06-24T13:09:21.753747+00:00",
                  "probability": [
                    0.6,
                    0.4
                  ]
                }
              ]
            },
            {
              "modelId": 38,
              "modelName": "Qwen3.5-397B-A17B",
              "predictions": [
                {
                  "time": "2026-05-20T03:20:23.593356+00:00",
                  "probability": [
                    0.24,
                    0.76
                  ]
                }
              ]
            },
            {
              "modelId": 54,
              "modelName": "Minimax-M2.7",
              "predictions": [
                {
                  "time": "2026-05-20T03:20:06.794767+00:00",
                  "probability": [
                    0.3,
                    0.7
                  ]
                },
                {
                  "time": "2026-06-11T03:21:39.591853+00:00",
                  "probability": [
                    0.75,
                    0.25
                  ]
                },
                {
                  "time": "2026-06-24T13:05:41.557007+00:00",
                  "probability": [
                    0.85,
                    0.15
                  ]
                }
              ]
            },
            {
              "modelId": 9,
              "modelName": "Deepseek-V3.2",
              "predictions": [
                {
                  "time": "2026-05-20T03:18:40.878912+00:00",
                  "probability": [
                    0.85,
                    0.15
                  ]
                }
              ]
            }
          ],
          "scores": [],
          "stats": {
            "totalModels": 12,
            "totalPredictions": 30
          }
        }
      }
    }
  }
}