[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"api:\u002Fnav:{}":3,"api:\u002Fbenchmarks\u002Fhle:{\"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":54,"categories":151,"tabs":153,"settings":155},{"id":106,"name":93,"slug":94,"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},6,"public",null,true,2158,"1.0000","0.0000","2026-10-28T09:47:38.906402+08:00",0,"Accuracy",1226628684,"2026-10-20",[119],{"id":120,"rank":54,"notes":108,"score":121,"access":122,"effort":123,"family":124,"is_tier":109,"metrics":125,"cost_usd":108,"hardware":108,"model_id":43,"provider":108,"settings":126,"data_date":134,"num_cases":110,"precision":108,"score_std":108,"categories":135,"harness_id":108,"is_primary":109,"model_name":12,"model_slug":13,"output_tps":108,"short_name":12,"variant_id":145,"effort_rank":47,"input_price":108,"num_samples":54,"vendor_name":8,"vendor_slug":9,"architecture":108,"benchmark_id":106,"evaluated_at":117,"harness_name":108,"harness_slug":108,"input_tokens":146,"output_price":108,"overall_rank":72,"ranked_count":54,"release_date":108,"ttft_seconds":108,"variant_name":147,"variant_slug":148,"vendor_order":114,"weights_repo":108,"missing_count":114,"output_tokens":149,"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":150},116,"0.3109","self_hosted","xhigh","Qwen3.8",{},{"top_p":127,"recipe":108,"harness":108,"judge_model":128,"num_samples":54,"temperature":129,"extra_settings":130,"scoring_method":108,"harness_version":108,"framework_version":132,"max_output_tokens":133},"1.000","Qwen3-235B-A22B","0.60",{"top_k":131},-1,"OPT 1478395",65536,"2026-10-22",[136,139,141,143],{"name":41,"rank":137,"slug":42,"score":138,"expected":43,"available":43,"ranked_count":72},3,"48.453230452674897000000000",{"name":45,"rank":72,"slug":46,"score":140,"expected":47,"available":47,"ranked_count":72},"45.520000000000000000000000",{"name":49,"rank":43,"slug":50,"score":142,"expected":43,"available":43,"ranked_count":72},"49.480000000000000000000000",{"name":52,"rank":137,"slug":53,"score":144,"expected":54,"available":54,"ranked_count":72},"41.500000000000000000000000",19,710766,"Qwen3.8-27B 自部署 xhigh","qwen3-8-27b-self-hosted-xhigh",77692867,64678674,[152],{"name":45,"slug":46},[154],{"name":93,"slug":94,"num_cases":110},{"extra_settings":156,"framework_version":158,"harness":159,"harness_version":160,"judge_model":161,"max_output_tokens":162,"num_samples":163,"recipe":164,"scoring_method":165,"temperature":166,"top_p":167},{"value":157,"varies":4,"label":108},{"top_k":131},{"value":132,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":128,"varies":4,"label":108},{"value":133,"varies":4,"label":108},{"value":54,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":108,"varies":4,"label":108},{"value":129,"varies":4,"label":108},{"value":127,"varies":4,"label":108}]