[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"api:\u002Fnav:{}":3,"api:\u002Fbenchmarks\u002Faa-lcr:{\"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":117,"rows":118,"total":72,"categories":222,"tabs":224,"settings":226},{"id":106,"name":81,"slug":82,"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},3,"public",null,true,100,"1.0000","0.0000","2026-10-28T09:47:38.906402+08:00",0,"Accuracy",3304837814,"2026-10-26",[119,152,177,199],{"id":120,"rank":54,"notes":108,"score":121,"access":122,"effort":123,"family":28,"is_tier":109,"metrics":124,"cost_usd":108,"hardware":108,"model_id":106,"provider":108,"settings":125,"data_date":133,"num_cases":110,"precision":108,"score_std":108,"categories":134,"harness_id":108,"is_primary":109,"model_name":28,"model_slug":29,"output_tps":108,"short_name":28,"variant_id":142,"effort_rank":143,"input_price":108,"num_samples":72,"vendor_name":24,"vendor_slug":25,"architecture":108,"benchmark_id":106,"evaluated_at":144,"harness_name":108,"harness_slug":108,"input_tokens":145,"output_price":108,"overall_rank":106,"ranked_count":72,"release_date":146,"ttft_seconds":108,"variant_name":147,"variant_slug":148,"vendor_order":114,"weights_repo":108,"missing_count":106,"output_tokens":149,"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":150,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":151},57,"0.8300","api","max",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":127,"num_samples":72,"temperature":128,"extra_settings":129,"scoring_method":108,"harness_version":108,"framework_version":131,"max_output_tokens":132},"1.000","Qwen3-235B-A22B","0.60",{"top_k":130},-1,"OPT 1478395",16384,"2026-10-24",[135,137,139,141],{"name":41,"rank":43,"slug":42,"score":136,"expected":43,"available":54,"ranked_count":72},"65.051903114186850000000000",{"name":45,"rank":43,"slug":46,"score":138,"expected":47,"available":72,"ranked_count":72},"67.050833333333332500000000",{"name":49,"rank":72,"slug":50,"score":140,"expected":43,"available":54,"ranked_count":72},"3.500000000000000000000000",{"name":52,"rank":72,"slug":53,"score":140,"expected":54,"available":54,"ranked_count":72},11,6,"2026-10-22",38064536,"2026-07-16","Kimi-K3 官方 API max","kimi-k3-api-max",445163,"147.42",51632013,{"id":153,"rank":43,"notes":108,"score":154,"access":155,"effort":156,"family":157,"is_tier":109,"metrics":158,"cost_usd":108,"hardware":108,"model_id":43,"provider":108,"settings":159,"data_date":144,"num_cases":110,"precision":108,"score_std":108,"categories":161,"harness_id":108,"is_primary":109,"model_name":12,"model_slug":13,"output_tps":108,"short_name":12,"variant_id":170,"effort_rank":47,"input_price":108,"num_samples":72,"vendor_name":8,"vendor_slug":9,"architecture":108,"benchmark_id":106,"evaluated_at":171,"harness_name":108,"harness_slug":108,"input_tokens":172,"output_price":108,"overall_rank":72,"ranked_count":72,"release_date":108,"ttft_seconds":108,"variant_name":173,"variant_slug":174,"vendor_order":114,"weights_repo":108,"missing_count":114,"output_tokens":175,"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":176},112,"0.8125","self_hosted","xhigh","Qwen3.8",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":127,"num_samples":72,"temperature":128,"extra_settings":160,"scoring_method":108,"harness_version":108,"framework_version":131,"max_output_tokens":132},{"top_k":130},[162,164,166,168],{"name":41,"rank":106,"slug":42,"score":163,"expected":43,"available":43,"ranked_count":72},"48.453230452674897000000000",{"name":45,"rank":72,"slug":46,"score":165,"expected":47,"available":47,"ranked_count":72},"45.520000000000000000000000",{"name":49,"rank":43,"slug":50,"score":167,"expected":43,"available":43,"ranked_count":72},"49.480000000000000000000000",{"name":52,"rank":106,"slug":53,"score":169,"expected":54,"available":54,"ranked_count":72},"41.500000000000000000000000",19,"2026-10-19",42647024,"Qwen3.8-27B 自部署 xhigh","qwen3-8-27b-self-hosted-xhigh",1065053,51697706,{"id":178,"rank":106,"notes":108,"score":179,"access":155,"effort":123,"family":180,"is_tier":109,"metrics":181,"cost_usd":108,"hardware":108,"model_id":72,"provider":108,"settings":182,"data_date":5,"num_cases":110,"precision":108,"score_std":108,"categories":184,"harness_id":108,"is_primary":109,"model_name":20,"model_slug":21,"output_tps":108,"short_name":20,"variant_id":106,"effort_rank":143,"input_price":108,"num_samples":72,"vendor_name":16,"vendor_slug":17,"architecture":108,"benchmark_id":106,"evaluated_at":117,"harness_name":108,"harness_slug":108,"input_tokens":192,"output_price":108,"overall_rank":54,"ranked_count":72,"release_date":193,"ttft_seconds":108,"variant_name":194,"variant_slug":195,"vendor_order":114,"weights_repo":108,"missing_count":106,"output_tokens":196,"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":197,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":198},12,"0.7550","DeepSeek-V4",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":127,"num_samples":72,"temperature":128,"extra_settings":183,"scoring_method":108,"harness_version":108,"framework_version":131,"max_output_tokens":132},{"top_k":130},[185,187,189,191],{"name":41,"rank":72,"slug":42,"score":186,"expected":43,"available":54,"ranked_count":72},"2.857142857142860000000000",{"name":45,"rank":54,"slug":46,"score":188,"expected":47,"available":72,"ranked_count":72},"69.724166666666667500000000",{"name":49,"rank":106,"slug":50,"score":190,"expected":43,"available":54,"ranked_count":72},"45.670000000000000000000000",{"name":52,"rank":43,"slug":53,"score":190,"expected":54,"available":54,"ranked_count":72},38197780,"2026-08-13","DeepSeek-V4-Pro 自部署 max","deepseek-v4-pro-self-hosted-max",1371986,"174.13",52355215,{"id":200,"rank":72,"notes":108,"score":201,"access":122,"effort":123,"family":36,"is_tier":109,"metrics":202,"cost_usd":108,"hardware":108,"model_id":54,"provider":108,"settings":203,"data_date":205,"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":143,"input_price":108,"num_samples":72,"vendor_name":32,"vendor_slug":33,"architecture":108,"benchmark_id":106,"evaluated_at":171,"harness_name":108,"harness_slug":108,"input_tokens":215,"output_price":108,"overall_rank":43,"ranked_count":72,"release_date":216,"ttft_seconds":108,"variant_name":217,"variant_slug":218,"vendor_order":114,"weights_repo":108,"missing_count":43,"output_tokens":219,"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":220,"inference_engine":108,"reasoning_tokens":108,"cost_per_case_usd":108,"trajectory_bundle_size_bytes":221},39,"0.6525",{},{"top_p":126,"recipe":108,"harness":108,"judge_model":127,"num_samples":72,"temperature":128,"extra_settings":204,"scoring_method":108,"harness_version":108,"framework_version":131,"max_output_tokens":132},{"top_k":130},"2026-10-23",[207,209,211,213],{"name":41,"rank":54,"slug":42,"score":208,"expected":43,"available":54,"ranked_count":72},"72.997032640949555000000000",{"name":45,"rank":106,"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",38927244,"2026-08-18","GLM-5.3 官方 API max","glm-5-3-api-max",1903602,"360.45",52569044,[223],{"name":45,"slug":46},[225],{"name":81,"slug":82,"num_cases":110},{"extra_settings":227,"framework_version":229,"harness":230,"harness_version":231,"judge_model":232,"max_output_tokens":233,"num_samples":234,"recipe":235,"scoring_method":236,"temperature":237,"top_p":238},{"value":228,"varies":4,"label":108},{"top_k":130},{"value":131,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":127,"varies":4,"label":108},{"value":132,"varies":4,"label":108},{"value":72,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":128,"varies":4,"label":108},{"value":126,"varies":4,"label":108}]