[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"api:\u002Fnav:{}":3,"api:\u002Fbenchmarks\u002Fmmlu-pro:{\"effort\":\"top\",\"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",{"benchmark":105,"updated_at":5,"rows":117,"total":72,"categories":223,"tabs":225,"settings":227},{"id":106,"name":96,"slug":97,"license":107,"version":108,"is_active":109,"num_cases":110,"parent_id":108,"score_max":111,"score_min":112,"created_at":113,"has_bundle":109,"is_agentic":4,"sort_order":114,"updated_at":113,"description":108,"metric_name":115,"homepage_url":108,"license_note":108,"scoring_method":108,"composite_weight":111,"higher_is_better":109,"bundle_size_bytes":116},7,"public",null,true,12032,"1.0000","0.0000","2026-10-28T09:47:38.906402+08:00",0,"Accuracy",1829320613,[118,151,174,199],{"id":119,"rank":54,"notes":108,"score":120,"access":121,"effort":122,"family":28,"is_tier":109,"metrics":123,"cost_usd":108,"hardware":108,"model_id":124,"provider":108,"settings":125,"data_date":132,"num_cases":110,"precision":108,"score_std":108,"categories":133,"harness_id":108,"is_primary":109,"model_name":28,"model_slug":29,"output_tps":108,"short_name":28,"variant_id":141,"effort_rank":142,"input_price":108,"num_samples":54,"vendor_name":24,"vendor_slug":25,"architecture":108,"benchmark_id":106,"evaluated_at":143,"harness_name":108,"harness_slug":108,"input_tokens":144,"output_price":108,"overall_rank":124,"ranked_count":72,"release_date":145,"ttft_seconds":108,"variant_name":146,"variant_slug":147,"vendor_order":114,"weights_repo":108,"missing_count":124,"output_tokens":148,"overall_score":30,"parent_run_id":108,"context_window":108,"is_primary_run":109,"total_cost_usd":108,"total_params_b":108,"weights_source":108,"active_params_b":108,"cache_hit_price":108,"cost_incomplete":109,"harness_version":108,"is_open_weights":109,"avg_case_seconds":149,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":150},60,"0.8877","api","max",{},3,{"top_p":126,"recipe":108,"harness":108,"judge_model":108,"num_samples":54,"temperature":127,"extra_settings":128,"scoring_method":108,"harness_version":108,"framework_version":130,"max_output_tokens":131},"1.000","0.60",{"top_k":129},-1,"OPT 1478395",65536,"2026-10-24",[134,136,138,140],{"name":41,"rank":43,"slug":42,"score":135,"expected":43,"available":54,"ranked_count":72},"65.051903114186850000000000",{"name":45,"rank":43,"slug":46,"score":137,"expected":47,"available":72,"ranked_count":72},"67.050833333333332500000000",{"name":49,"rank":72,"slug":50,"score":139,"expected":43,"available":54,"ranked_count":72},"3.500000000000000000000000",{"name":52,"rank":72,"slug":53,"score":139,"expected":54,"available":54,"ranked_count":72},11,6,"2026-10-23",3433876,"2026-07-16","Kimi-K3 官方 API max","kimi-k3-api-max",14673875,"107.58",21102835,{"id":152,"rank":43,"notes":108,"score":153,"access":154,"effort":122,"family":155,"is_tier":109,"metrics":156,"cost_usd":108,"hardware":108,"model_id":72,"provider":108,"settings":157,"data_date":5,"num_cases":110,"precision":108,"score_std":108,"categories":159,"harness_id":108,"is_primary":109,"model_name":20,"model_slug":21,"output_tps":108,"short_name":20,"variant_id":124,"effort_rank":142,"input_price":108,"num_samples":54,"vendor_name":16,"vendor_slug":17,"architecture":108,"benchmark_id":106,"evaluated_at":5,"harness_name":108,"harness_slug":108,"input_tokens":167,"output_price":108,"overall_rank":54,"ranked_count":72,"release_date":168,"ttft_seconds":108,"variant_name":169,"variant_slug":170,"vendor_order":114,"weights_repo":108,"missing_count":124,"output_tokens":171,"overall_score":22,"parent_run_id":108,"context_window":108,"is_primary_run":109,"total_cost_usd":108,"total_params_b":108,"weights_source":108,"active_params_b":108,"cache_hit_price":108,"cost_incomplete":109,"harness_version":108,"is_open_weights":109,"avg_case_seconds":172,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":173},15,"0.8713","self_hosted","DeepSeek-V4",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":108,"num_samples":54,"temperature":127,"extra_settings":158,"scoring_method":108,"harness_version":108,"framework_version":130,"max_output_tokens":131},{"top_k":129},[160,162,164,166],{"name":41,"rank":72,"slug":42,"score":161,"expected":43,"available":54,"ranked_count":72},"2.857142857142860000000000",{"name":45,"rank":54,"slug":46,"score":163,"expected":47,"available":72,"ranked_count":72},"69.724166666666667500000000",{"name":49,"rank":124,"slug":50,"score":165,"expected":43,"available":54,"ranked_count":72},"45.670000000000000000000000",{"name":52,"rank":43,"slug":53,"score":165,"expected":54,"available":54,"ranked_count":72},3552394,"2026-08-13","DeepSeek-V4-Pro 自部署 max","deepseek-v4-pro-self-hosted-max",47930546,"138.95",45236996,{"id":175,"rank":124,"notes":108,"score":176,"access":154,"effort":177,"family":178,"is_tier":109,"metrics":179,"cost_usd":108,"hardware":108,"model_id":43,"provider":108,"settings":180,"data_date":182,"num_cases":110,"precision":108,"score_std":108,"categories":183,"harness_id":108,"is_primary":109,"model_name":12,"model_slug":13,"output_tps":108,"short_name":12,"variant_id":192,"effort_rank":47,"input_price":108,"num_samples":54,"vendor_name":8,"vendor_slug":9,"architecture":108,"benchmark_id":106,"evaluated_at":193,"harness_name":108,"harness_slug":108,"input_tokens":194,"output_price":108,"overall_rank":72,"ranked_count":72,"release_date":108,"ttft_seconds":108,"variant_name":195,"variant_slug":196,"vendor_order":114,"weights_repo":108,"missing_count":114,"output_tokens":197,"overall_score":14,"parent_run_id":108,"context_window":108,"is_primary_run":109,"total_cost_usd":108,"total_params_b":108,"weights_source":108,"active_params_b":108,"cache_hit_price":108,"cost_incomplete":109,"harness_version":108,"is_open_weights":109,"avg_case_seconds":108,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":198},117,"0.8516","xhigh","Qwen3.8",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":108,"num_samples":54,"temperature":127,"extra_settings":181,"scoring_method":108,"harness_version":108,"framework_version":130,"max_output_tokens":131},{"top_k":129},"2026-10-22",[184,186,188,190],{"name":41,"rank":124,"slug":42,"score":185,"expected":43,"available":43,"ranked_count":72},"48.453230452674897000000000",{"name":45,"rank":72,"slug":46,"score":187,"expected":47,"available":47,"ranked_count":72},"45.520000000000000000000000",{"name":49,"rank":43,"slug":50,"score":189,"expected":43,"available":43,"ranked_count":72},"49.480000000000000000000000",{"name":52,"rank":124,"slug":53,"score":191,"expected":54,"available":54,"ranked_count":72},"41.500000000000000000000000",19,"2026-10-20",3238380,"Qwen3.8-27B 自部署 xhigh","qwen3-8-27b-self-hosted-xhigh",54078637,46543663,{"id":200,"rank":72,"notes":108,"score":201,"access":121,"effort":122,"family":36,"is_tier":109,"metrics":202,"cost_usd":108,"hardware":108,"model_id":54,"provider":108,"settings":203,"data_date":143,"num_cases":110,"precision":108,"score_std":108,"categories":206,"harness_id":108,"is_primary":109,"model_name":36,"model_slug":37,"output_tps":108,"short_name":36,"variant_id":73,"effort_rank":142,"input_price":108,"num_samples":54,"vendor_name":32,"vendor_slug":33,"architecture":108,"benchmark_id":106,"evaluated_at":215,"harness_name":108,"harness_slug":108,"input_tokens":216,"output_price":108,"overall_rank":43,"ranked_count":72,"release_date":217,"ttft_seconds":108,"variant_name":218,"variant_slug":219,"vendor_order":114,"weights_repo":108,"missing_count":43,"output_tokens":220,"overall_score":38,"parent_run_id":108,"context_window":108,"is_primary_run":109,"total_cost_usd":108,"total_params_b":108,"weights_source":108,"active_params_b":108,"cache_hit_price":108,"cost_incomplete":109,"harness_version":108,"is_open_weights":109,"avg_case_seconds":221,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":222},42,"0.8507",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":108,"num_samples":54,"temperature":127,"extra_settings":204,"scoring_method":108,"harness_version":108,"framework_version":130,"max_output_tokens":205},{"top_k":129},32768,[207,209,211,213],{"name":41,"rank":54,"slug":42,"score":208,"expected":43,"available":54,"ranked_count":72},"72.997032640949555000000000",{"name":45,"rank":124,"slug":46,"score":210,"expected":47,"available":72,"ranked_count":72},"59.363333333333333750000000",{"name":49,"rank":54,"slug":50,"score":212,"expected":43,"available":43,"ranked_count":72},"54.225000000000000000000000",{"name":52,"rank":54,"slug":53,"score":214,"expected":54,"available":54,"ranked_count":72},"52.500000000000000000000000","2026-10-16",2583897,"2026-08-18","GLM-5.3 官方 API max","glm-5-3-api-max",50800608,"227.07",46257854,[224],{"name":45,"slug":46},[226],{"name":96,"slug":97,"num_cases":110},{"extra_settings":228,"framework_version":230,"harness":231,"harness_version":232,"judge_model":233,"max_output_tokens":234,"num_samples":236,"recipe":237,"scoring_method":238,"temperature":239,"top_p":240},{"value":229,"varies":4,"label":108},{"top_k":129},{"value":130,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":109,"label":235},"各记录不同",{"value":54,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":127,"varies":4,"label":108},{"value":126,"varies":4,"label":108}]