[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"api:\u002Fnav:{}":3,"api:\u002Fbenchmarks\u002Fcategory\u002Fagentic:{\"q\":\"\",\"sort\":\"updated\",\"limit\":100,\"page\":1}":104},{"login_code_required":4,"data_date":5,"vendors":6,"categories":39,"topics":55,"model_count":72,"benchmark_count":73,"models":74,"benchmarks":79},false,"2026-10-27",[7,15,23,31],{"name":8,"slug":9,"models":10},"阿里巴巴","alibaba",[11],{"name":12,"slug":13,"short_name":12,"overall_score":14},"Qwen3.8-27B","qwen3-8-27b","47.576875000000000000000000",{"name":16,"slug":17,"models":18},"深度求索","deepseek",[19],{"name":20,"slug":21,"short_name":20,"overall_score":22},"DeepSeek-V4-Pro","deepseek-v4-pro","65.030000000000000000000000",{"name":24,"slug":25,"models":26},"月之暗面","moonshot",[27],{"name":28,"slug":29,"short_name":28,"overall_score":30},"Kimi-K3","kimi-k3","57.474000000000000000000000",{"name":32,"slug":33,"models":34},"智谱","zhipu",[35],{"name":36,"slug":37,"short_name":36,"overall_score":38},"GLM-5.3","glm-5-3","61.067500000000000000000000",[40,44,48,51],{"name":41,"slug":42,"benchmark_count":43},"体验感","experience",2,{"name":45,"slug":46,"benchmark_count":47},"认知能力","cognition",5,{"name":49,"slug":50,"benchmark_count":43},"Agentic 能力","agentic",{"name":52,"slug":53,"benchmark_count":54},"应用领域","domain",1,[56,60,64,68],{"slug":57,"name":58,"question":59},"quant","量化 vs 原精度","量化之后能力掉多少",{"slug":61,"name":62,"question":63},"selfhost","自部署 vs 官方 API","同一份权重，自己部署和调官方 API，分数一致吗",{"slug":65,"name":66,"question":67},"size","同家族尺寸","同一代模型，小尺寸比旗舰差多少",{"slug":69,"name":70,"question":71},"harness","多 Harness","同一个模型换一个 harness，agentic 分数变多少",4,8,[75,76,77,78],{"name":20,"slug":21},{"name":36,"slug":37},{"name":28,"slug":29},{"name":12,"slug":13},[80,83,86,89,92,95,98,101],{"name":81,"slug":82},"AA-LCR","aa-lcr",{"name":84,"slug":85},"AA-Omniscience","aa-omniscience",{"name":87,"slug":88},"AutomationBench","automationbench",{"name":90,"slug":91},"CL-bench","cl-bench",{"name":93,"slug":94},"HLE","hle",{"name":96,"slug":97},"MMLU-Pro","mmlu-pro",{"name":99,"slug":100},"SimpleQA Verified","simpleqa-verified",{"name":102,"slug":103},"SWE-bench Pro","swe-bench-pro",{"category":105,"total":43,"benchmarks":109},{"id":106,"name":49,"slug":50,"created_at":107,"sort_order":43,"updated_at":107,"description":108},3,"2026-10-28T09:47:38.906402+08:00","任务规划与分解、工具调用和记忆管理",[110,130],{"id":43,"best":111,"name":87,"slug":88,"license":122,"num_cases":123,"categories":124,"is_agentic":127,"updated_at":128,"description":116,"metric_name":129},{"name":36,"score":112,"run_id":113,"settings":114,"model_slug":37,"vendor_slug":33},"0.5250",41,{"top_p":115,"recipe":116,"harness":116,"judge_model":116,"num_samples":54,"temperature":117,"extra_settings":118,"scoring_method":116,"harness_version":116,"framework_version":120,"max_output_tokens":121},"1.000",null,"0.00",{"top_k":119},-1,"OPT 1478395",32768,"public",600,[125,126],{"name":49,"slug":50},{"name":52,"slug":53},true,"2026-10-26","Pass Rate",{"id":131,"best":132,"name":102,"slug":103,"license":122,"num_cases":139,"categories":140,"is_agentic":127,"updated_at":142,"description":116,"metric_name":143},9,{"name":12,"score":133,"run_id":134,"settings":135,"model_slug":13,"vendor_slug":9},"0.5746",120,{"top_p":116,"recipe":116,"harness":136,"judge_model":116,"num_samples":54,"temperature":116,"extra_settings":137,"scoring_method":116,"harness_version":116,"framework_version":138,"max_output_tokens":116},"Codex CLI",{},"OPT 06dca09",731,[141],{"name":49,"slug":50},"2026-10-23","Resolved Rate"]