{"configurations":[{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.510324483775811,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.620967741935484,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.166666666666667,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.071135445543},"score_percent":{"formula":"score.value * 100","value":43.26529641259874}},"effort":"default","effort_raw":null,"harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":false,"version":"2.1.224; 2.1.261"},"id":"alibaba:qwen3-8-max:claude-code:default","model":{"aliases":["Qwen3.8 Max"],"api_id":null,"api_id_source_url":null,"display_name":"Qwen3.8 Max","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"alibaba","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.43265296412598736},"source_configuration_id":"cd4c865dc2f539b650d741fd2ebf7f2d","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9507113544554301,"cost_usd":3.478965577837015,"refusal_rate":0,"runtime_seconds":3784.416073707372,"tokens":8706893.798112892}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":null},"id":"alibaba:qwen3-8-max:opencode:default","model":{"aliases":["Qwen3.8 Max"],"api_id":null,"api_id_source_url":null,"display_name":"Qwen3.8 Max","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"alibaba","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"qwen3-8-max-opencode-default-unspecified","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"compound_configuration","components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.631268436578171,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.658602150537635,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.560606060606061,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.84900963367991},"score_percent":{"formula":"score.value * 100","value":61.68255492406223}},"effort":"default","effort_raw":"Claude Fable 5.1 XHigh + SWE-2 Medium","harness":{"name":"Devin Fusion CLI","slug":"devin-fusion-cli","vendor_first_party":false,"version":"3000.10.7"},"id":"anthropic-cognition:claude-fable-5-1-xhigh-swe-2-medium:devin-fusion-cli:compound-xhigh-medium","model":{"aliases":["Claude Fable 5.1 XHigh + SWE-2 Medium"],"api_id":null,"api_id_source_url":null,"display_name":"Claude Fable 5.1 XHigh + SWE-2 Medium","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"anthropic-cognition","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6168255492406223},"source_configuration_id":"34970c963bdfd318e46205514d46ea4f","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9584900963367992,"cost_usd":7.897126985217273,"refusal_rate":0.07076659294048016,"runtime_seconds":2146.9920896589642,"tokens":9691548.173817372}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"high","effort_raw":"high","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":null},"id":"anthropic:fable-5-1:claude-code:high","model":{"aliases":["Fable 5.1"],"api_id":"claude-fable-5-1","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Fable 5.1","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-fable-5.1","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.25,"input_per_mtok_usd":10,"output_per_mtok_usd":50,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"fable-5-1-claude-code-high","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.64306784660767,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.647849462365591,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.575757575757576,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":91.59555838702424},"score_percent":{"formula":"score.value * 100","value":62.22249615769457}},"effort":"max","effort_raw":"max","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.259; 2.1.263"},"id":"anthropic:fable-5-1:claude-code:max","model":{"aliases":["Fable 5.1"],"api_id":"claude-fable-5-1","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Fable 5.1","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-fable-5.1","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.25,"input_per_mtok_usd":10,"output_per_mtok_usd":50,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6222249615769457},"source_configuration_id":"29a1cdad0d140781d05b9b4746cfdd0d","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9159555838702423,"cost_usd":12.388092375137534,"refusal_rate":0.08835668513087869,"runtime_seconds":2089.900125412543,"tokens":5715253.85918592}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":null},"id":"anthropic:fable-5-1:claude-code:xhigh","model":{"aliases":["Fable 5.1"],"api_id":"claude-fable-5-1","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Fable 5.1","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-fable-5.1","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.25,"input_per_mtok_usd":10,"output_per_mtok_usd":50,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"fable-5-1-claude-code-xhigh","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.463126843657817,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.46236559139785,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.126262626262626,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.61309602762921},"score_percent":{"formula":"score.value * 100","value":35.0585020439431}},"effort":"high","effort_raw":"high","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.289"},"id":"anthropic:haiku-5-5:claude-code:high","model":{"aliases":["Haiku 5.5"],"api_id":"claude-haiku-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Haiku 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-haiku-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard_up_to_100k_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.350585020439431},"source_configuration_id":"da2f03dcda6604f53d83189400698b0a","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9561309602762921,"cost_usd":0.36467934237073724,"refusal_rate":0.005050505050505051,"runtime_seconds":1106.807743674368,"tokens":5076190.689768974}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.454277286135693,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.330645161290323,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.0606060606060606,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":94.78942152323016},"score_percent":{"formula":"score.value * 100","value":28.18428360106922}},"effort":"low","effort_raw":"low","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.289"},"id":"anthropic:haiku-5-5:claude-code:low","model":{"aliases":["Haiku 5.5"],"api_id":"claude-haiku-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Haiku 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-haiku-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard_up_to_100k_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.2818428360106922},"source_configuration_id":"b4a228ec941f485bc4e96489bb773dea","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9478942152323017,"cost_usd":0.14324741378437839,"refusal_rate":0.0033670033670033673,"runtime_seconds":693.5464620462047,"tokens":2412388.3531353152}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.460176991150442,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.336021505376344,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.297979797979798,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":89.18200950779239},"score_percent":{"formula":"score.value * 100","value":36.472609816886134}},"effort":"max","effort_raw":"max","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.289"},"id":"anthropic:haiku-5-5:claude-code:max","model":{"aliases":["Haiku 5.5"],"api_id":"claude-haiku-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Haiku 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-haiku-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard_up_to_100k_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.3647260981688613},"source_configuration_id":"69061178d1f87a145d332b38f566b7d0","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.8918200950779239,"cost_usd":2.5803409612582775,"refusal_rate":0.005050505050505051,"runtime_seconds":5225.859279249448,"tokens":13769015.177704195}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.474926253687316,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.416666666666667,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.116161616161616,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.547252331787},"score_percent":{"formula":"score.value * 100","value":33.5918178838533}},"effort":"medium","effort_raw":"medium","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.289"},"id":"anthropic:haiku-5-5:claude-code:medium","model":{"aliases":["Haiku 5.5"],"api_id":"claude-haiku-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Haiku 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-haiku-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard_up_to_100k_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.33591817883853303},"source_configuration_id":"430db5363624407d35b63d80ec863ab6","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.95547252331787,"cost_usd":0.22638840947194713,"refusal_rate":0.006734006734006735,"runtime_seconds":837.3769053905384,"tokens":3480808.845984597}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.489675516224189,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.516129032258065,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.237373737373737,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.02433850815568},"score_percent":{"formula":"score.value * 100","value":41.439276195199696}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.289"},"id":"anthropic:haiku-5-5:claude-code:xhigh","model":{"aliases":["Haiku 5.5"],"api_id":"claude-haiku-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Haiku 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-haiku-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard_up_to_100k_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.414392761951997},"source_configuration_id":"e52072ed5528ec319d2f8030a2cf63a2","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9502433850815568,"cost_usd":0.6104935163943895,"refusal_rate":0.001692047377326565,"runtime_seconds":1552.9344400440023,"tokens":7080866.402640261}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"high","effort_raw":"high","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":null},"id":"anthropic:opus-5-5:claude-code:high","model":{"aliases":["Opus 5.5"],"api_id":"claude-opus-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Opus 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-opus-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-18T07:06:28.741Z","cache_read_per_mtok_usd":0.2,"input_per_mtok_usd":4,"output_per_mtok_usd":20,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"opus-5-5-claude-code-high","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.684365781710915,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.663978494623656,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.631313131313131,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.00106599458849},"score_percent":{"formula":"score.value * 100","value":65.9885802549234}},"effort":"max","effort_raw":"max","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:opus-5-5:claude-code:max","model":{"aliases":["Opus 5.5"],"api_id":"claude-opus-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Opus 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-opus-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-18T07:06:28.741Z","cache_read_per_mtok_usd":0.2,"input_per_mtok_usd":4,"output_per_mtok_usd":20,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.659885802549234},"source_configuration_id":"ab5074e404be48bbd4ac3c488ad9a3e2","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9500106599458848,"cost_usd":13.036383418426828,"refusal_rate":0.08896559082114319,"runtime_seconds":3867.0923410341024,"tokens":15551383.220022004}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":null},"id":"anthropic:opus-5-5:claude-code:xhigh","model":{"aliases":["Opus 5.5"],"api_id":"claude-opus-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Opus 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-opus-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-18T07:06:28.741Z","cache_read_per_mtok_usd":0.2,"input_per_mtok_usd":4,"output_per_mtok_usd":20,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"opus-5-5-claude-code-xhigh","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.625368731563422,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.620967741935484,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.545454545454545,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.98258997320075},"score_percent":{"formula":"score.value * 100","value":59.726367298448366}},"effort":"max","effort_raw":"max","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.218; 2.1.263"},"id":"anthropic:opus-5:claude-code:max","model":{"aliases":["Opus 5"],"api_id":null,"api_id_source_url":null,"display_name":"Opus 5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-opus-5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.5,"input_per_mtok_usd":5,"output_per_mtok_usd":25,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5972636729844837},"source_configuration_id":"409518818f1e9ad21995d70631ab05b6","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9698258997320075,"cost_usd":10.789154699257413,"refusal_rate":0.005076142131979696,"runtime_seconds":2515.540235973597,"tokens":11374850.488998907}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.666666666666667,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.564516129032258,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.419191919191919,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":94.77934370079522},"score_percent":{"formula":"score.value * 100","value":55.01249049636148}},"effort":"high","effort_raw":"high","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:sonnet-5-5:claude-code:high","model":{"aliases":["Sonnet 5.5"],"api_id":"claude-sonnet-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Sonnet 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-sonnet-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5501249049636148},"source_configuration_id":"33e80f383dc644527161435b076798ad","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9477934370079523,"cost_usd":1.235497127172719,"refusal_rate":0.031144781144781142,"runtime_seconds":735.1809878987912,"tokens":2652380.565456549}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.619469026548673,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.389784946236559,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.252525252525253,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.07814644010347},"score_percent":{"formula":"score.value * 100","value":42.0593075103495}},"effort":"low","effort_raw":"low","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:sonnet-5-5:claude-code:low","model":{"aliases":["Sonnet 5.5"],"api_id":"claude-sonnet-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Sonnet 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-sonnet-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.420593075103495},"source_configuration_id":"e9f9d3ae5b03aa060114dd4b3275d67d","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9307814644010347,"cost_usd":0.48311011474147336,"refusal_rate":0.018491365265558814,"runtime_seconds":380.85617931793183,"tokens":977284.7700770069}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.71976401179941,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.669354838709677,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.661616161616162,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.06214943317096},"score_percent":{"formula":"score.value * 100","value":68.35783373750832}},"effort":"max","effort_raw":"max","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:sonnet-5-5:claude-code:max","model":{"aliases":["Sonnet 5.5"],"api_id":"claude-sonnet-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Sonnet 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-sonnet-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6835783373750831},"source_configuration_id":"a2c87c062f3cef73d8525e7578f14742","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9506214943317096,"cost_usd":14.190800796369649,"refusal_rate":0.045047246660149885,"runtime_seconds":5244.965372937293,"tokens":27742337.233773403}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.654867256637168,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.448924731182796,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.272727272727273,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.41009883549258},"score_percent":{"formula":"score.value * 100","value":45.88397535157457}},"effort":"medium","effort_raw":"medium","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:sonnet-5-5:claude-code:medium","model":{"aliases":["Sonnet 5.5"],"api_id":"claude-sonnet-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Sonnet 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-sonnet-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4588397535157457},"source_configuration_id":"4d6aa9e9c8df931d54dba5956d94973c","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9341009883549257,"cost_usd":0.6187507753025301,"refusal_rate":0.018599978277397632,"runtime_seconds":510.7721826182621,"tokens":1253351.4928492843}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.684365781710914,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.620967741935484,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.580808080808081,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.74714756466342},"score_percent":{"formula":"score.value * 100","value":62.8713868151493}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":true,"version":"2.1.280"},"id":"anthropic:sonnet-5-5:claude-code:xhigh","model":{"aliases":["Sonnet 5.5"],"api_id":"claude-sonnet-5-5","api_id_source_url":"https://platform.claude.com/docs/en/models/overview","display_name":"Sonnet 5.5","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"anthropic/claude-sonnet-5.5","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"anthropic","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://platform.claude.com/docs/en/about-claude/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.628713868151493},"source_configuration_id":"666605c66ad569c503ce53c01682e285","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9574714756466342,"cost_usd":3.3333865348734886,"refusal_rate":0.032828282828282825,"runtime_seconds":1619.0134708470848,"tokens":7087678.3960396005}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"compound_configuration","components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.725663716814159,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.696236559139785,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.621212121212121,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":87.12080457734012},"score_percent":{"formula":"score.value * 100","value":68.10374657220217}},"effort":"default","effort_raw":"Claude Opus 5.5 Max + GPT-6 Luna High","harness":{"name":"Devin Fusion CLI","slug":"devin-fusion-cli","vendor_first_party":true,"version":"3000.11.34 (DeepSWE v1.1); 3000.11.31 (SWE-Atlas-QnA, Terminal-Bench v4)"},"id":"cognition:claude-opus-5-5-max-gpt-6-luna-high:devin-fusion-cli:compound-max-high","model":{"aliases":["Claude Opus 5.5 Max + GPT-6 Luna High"],"api_id":null,"api_id_source_url":null,"display_name":"Claude Opus 5.5 Max + GPT-6 Luna High","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"cognition","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6810374657220217},"source_configuration_id":"9bafad3e22caac44923f3df721644e06","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.8712080457734013,"cost_usd":9.00923788524477,"refusal_rate":0,"runtime_seconds":3289.9155137513735,"tokens":12806701.57480748}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"Unverified / no current AA configuration","slug":"unverified-no-current-aa-configuration","vendor_first_party":false,"version":null},"id":"deepseek:deepseek-v4-1-flash-1017:unverified-no-current-aa-configuration:default","model":{"aliases":["DeepSeek-V4.1-Flash-1017"],"api_id":null,"api_id_source_url":null,"display_name":"DeepSeek-V4.1-Flash-1017","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"deepseek","weights":"open","weights_url":"https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017"},"pricing":null,"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"deepseek-v4-1-flash-1017-unverified-harness-default-unspecified","source_retrieved_at":"2026-10-18T07:06:28.741Z","source_url":"https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.542772861356932,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.513440860215054,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.106060606060606,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":98.69993314104538},"score_percent":{"formula":"score.value * 100","value":38.74247758775307}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":false,"version":"0.146.0; 0.153.2"},"id":"deepseek:deepseek-v4-flash-0731:codex:max","model":{"aliases":["DeepSeek V4 Flash 0731"],"api_id":null,"api_id_source_url":null,"display_name":"DeepSeek V4 Flash 0731","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"deepseek","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.3874247758775307},"source_configuration_id":"28ef65135f08f035dc93a97d324fea73","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9869993314104538,"cost_usd":0.08506419165060494,"refusal_rate":0,"runtime_seconds":1097.0404400440034,"tokens":15350225.299229927}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.572271386430678,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.618279569892473,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.101010101010101,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":98.98485564269163},"score_percent":{"formula":"score.value * 100","value":43.05203524444174}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":false,"version":"0.147.0; 0.153.2"},"id":"deepseek:deepseek-v4-pro-0813:codex:max","model":{"aliases":["DeepSeek V4 Pro 0813"],"api_id":null,"api_id_source_url":null,"display_name":"DeepSeek V4 Pro 0813","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"deepseek/deepseek-v4-pro-0813","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"deepseek","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4305203524444174},"source_configuration_id":"542590ecc1aa12a718d158565d464daa","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9898485564269163,"cost_usd":0.23804610720792058,"refusal_rate":0,"runtime_seconds":2417.5500099009905,"tokens":24180108.028052814}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.657817109144543,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.451612903225807,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.146464646464646,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":86.97428234581179},"score_percent":{"formula":"score.value * 100","value":41.86315529449987}},"effort":"high","effort_raw":"high","harness":{"name":"Antigravity SDK","slug":"antigravity-sdk","vendor_first_party":true,"version":"0.1.12; 0.1.16"},"id":"google:gemini-3-8-flash:antigravity-sdk:high","model":{"aliases":["Gemini 3.8 Flash"],"api_id":null,"api_id_source_url":null,"display_name":"Gemini 3.8 Flash","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"google/gemini-3.8-flash","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"google","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4186315529449987},"source_configuration_id":"d1003684d8f29b95a3b14d2a7722221a","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.8697428234581179,"cost_usd":2.4673922004125437,"refusal_rate":0,"runtime_seconds":702.7110253025301,"tokens":13511073.155115515}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.787610619469027,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.564516129032258,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.560606060606061,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":87.00000000000004},"score_percent":{"formula":"score.value * 100","value":63.7577603035782}},"effort":"default","effort_raw":null,"harness":{"name":"Antigravity CLI","slug":"antigravity-cli","vendor_first_party":true,"version":"1.2.10"},"id":"google:gemini-4-argon:antigravity-cli:default","model":{"aliases":["Gemini 4 Argon"],"api_id":null,"api_id_source_url":null,"display_name":"Gemini 4 Argon","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"google","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-15T07:02:47.644Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://ai.google.dev/gemini-api/docs/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.637577603035782},"source_configuration_id":"ef63fafa9b4a635522183c879ae8435b","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.8700000000000004,"cost_usd":5.8439053894448145,"refusal_rate":0,"runtime_seconds":2070.1686032008843,"tokens":13733793.781456953}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.716814159292036,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.594086021505376,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.318181818181818,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":97.12826195905684},"score_percent":{"formula":"score.value * 100","value":54.30273329930766}},"effort":"max","effort_raw":"max","harness":{"name":"Muse Code","slug":"muse-code","vendor_first_party":true,"version":"1.0.3-R2198.1"},"id":"meta:muse-spark-1-3:muse-code:max","model":{"aliases":["Muse Spark 1.3"],"api_id":null,"api_id_source_url":null,"display_name":"Muse Spark 1.3","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"meta/muse-spark-1.3","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"meta","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5430273329930766},"source_configuration_id":"f2de07373c3626f1b9269a6ecad7804c","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9712826195905685,"cost_usd":3.977504096947204,"refusal_rate":0,"runtime_seconds":1106.8533531353135,"tokens":16709444.57370737}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.731563421828909,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.545698924731183,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.171717171717172,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.61832360411204},"score_percent":{"formula":"score.value * 100","value":48.2993172759088}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Muse Code","slug":"muse-code","vendor_first_party":true,"version":"1.0.2-R2040.1"},"id":"meta:muse-spark-1-3:muse-code:xhigh","model":{"aliases":["Muse Spark 1.3"],"api_id":null,"api_id_source_url":null,"display_name":"Muse Spark 1.3","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"meta/muse-spark-1.3","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"meta","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.482993172759088},"source_configuration_id":"cd60470c35fd626a116af80fd5915dfa","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9661832360411203,"cost_usd":3.467754041474145,"refusal_rate":0,"runtime_seconds":1098.202911991198,"tokens":16237094.352035202}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.601769911504425,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.577956989247312,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.0555555555555556,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.9270887783539},"score_percent":{"formula":"score.value * 100","value":41.176081876909755}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Muse Code","slug":"muse-code","vendor_first_party":true,"version":"1.4.2-R4684.1"},"id":"meta:muse-spark-open:muse-code:xhigh","model":{"aliases":["Muse Spark Open"],"api_id":null,"api_id_source_url":null,"display_name":"Muse Spark Open","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"meta","weights":"open","weights_url":"https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B"},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.41176081876909754},"source_configuration_id":"2f37603a93bf2cdddfdb7fa9ccce4803","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9692708877835391,"cost_usd":3.3259927397139704,"refusal_rate":0,"runtime_seconds":1513.8331193619354,"tokens":16314434.26072608}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.769911504424779,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.68010752688172,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.590909090909091,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.37709553156532},"score_percent":{"formula":"score.value * 100","value":68.03093740718634}},"effort":"max","effort_raw":"max","harness":{"name":"Muse Code","slug":"muse-code","vendor_first_party":true,"version":"1.4.5-R5426.1"},"id":"meta:muse-wonder-full:muse-code:max","model":{"aliases":["Muse Wonder Full"],"api_id":null,"api_id_source_url":null,"display_name":"Muse Wonder Full","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"meta","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6803093740718634},"source_configuration_id":"60d5b66f39c9e9aacacde1855d69e211","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9637709553156532,"cost_usd":null,"refusal_rate":0,"runtime_seconds":1725.8832958057408,"tokens":15002315.66997792}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.75811209439528,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.672043010752688,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.474747474747475,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.57352125258251},"score_percent":{"formula":"score.value * 100","value":63.49675266318143}},"effort":"max","effort_raw":"max","harness":{"name":"Muse Code","slug":"muse-code","vendor_first_party":true,"version":"1.4.5-R5426.1"},"id":"meta:muse-wonder:muse-code:max","model":{"aliases":["Muse Wonder"],"api_id":null,"api_id_source_url":null,"display_name":"Muse Wonder","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"meta","weights":"unknown","weights_url":null},"pricing":{"as_of":"2026-10-13T07:15:11.407Z","cache_read_per_mtok_usd":0.2,"input_per_mtok_usd":4,"output_per_mtok_usd":20,"source_url":"https://dev.meta.ai/models/muse-wonder","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6349675266318143},"source_configuration_id":"61df0222d90b30e25d4ff8b08d0c0aca","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.935735212525825,"cost_usd":8.052232502530256,"refusal_rate":0,"runtime_seconds":1346.6521254125387,"tokens":15016299.67931794}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":null},"id":"mistral:mistral-large-4:opencode:default","model":{"aliases":["Mistral Large 4"],"api_id":null,"api_id_source_url":null,"display_name":"Mistral Large 4","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"mistralai/mistral-large-4-0","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"mistral","weights":"unknown","weights_url":null},"pricing":{"as_of":"2026-10-18T07:06:28.741Z","cache_read_per_mtok_usd":0.07,"input_per_mtok_usd":0.68,"output_per_mtok_usd":2.09,"source_url":"https://docs.mistral.ai/models/mistral-large-4-0","tier":"preview_sale"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"mistral-large-4-opencode-default-unspecified","source_retrieved_at":"2026-10-18T07:06:28.741Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.684365781710915,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.661290322580645,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.212121212121212,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":95.00000000000017},"score_percent":{"formula":"score.value * 100","value":51.9259105470924}},"effort":"default","effort_raw":null,"harness":{"name":"Kimi Code CLI","slug":"kimi-code-cli","vendor_first_party":true,"version":"0.26.0; 0.36.0; 0.36.1"},"id":"moonshot:kimi-k3:kimi-code-cli:default","model":{"aliases":["Kimi K3"],"api_id":null,"api_id_source_url":null,"display_name":"Kimi K3","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"moonshot","weights":"open","weights_url":"https://huggingface.co/moonshotai/Kimi-K3"},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.3,"input_per_mtok_usd":3,"output_per_mtok_usd":15,"source_url":"https://www.kimi.ai/blog/kimi-k3","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.519259105470924},"source_configuration_id":"55b029d6971b4d378335bb168eb98b3b","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9500000000000017,"cost_usd":5.047195443300332,"refusal_rate":0,"runtime_seconds":3689.3441947194724,"tokens":8481121.533553353}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"Unverified / no current AA configuration","slug":"unverified-no-current-aa-configuration","vendor_first_party":false,"version":null},"id":"nvidia:nvidia-nemotron-labs-dory-1-0-4b-bf16:unverified-no-current-aa-configuration:default","model":{"aliases":["NVIDIA Nemotron Labs Dory 1.0 4B BF16"],"api_id":null,"api_id_source_url":null,"display_name":"NVIDIA Nemotron Labs Dory 1.0 4B BF16","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"nvidia","weights":"open","weights_url":"https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16"},"pricing":null,"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"nvidia-dory-1-0-4b-bf16-unverified-harness-default-unspecified","source_retrieved_at":"2026-10-18T07:06:28.741Z","source_url":"https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"compound_configuration","components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.672566371681416,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.594086021505376,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.5,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":92.0061132172241},"score_percent":{"formula":"score.value * 100","value":58.8884131062264}},"effort":"default","effort_raw":"GPT-6 Astra XHigh + SWE-2 Medium","harness":{"name":"Devin Fusion CLI","slug":"devin-fusion-cli","vendor_first_party":false,"version":"3000.10.7"},"id":"openai-cognition:gpt-6-astra-xhigh-swe-2-medium:devin-fusion-cli:compound-xhigh-medium","model":{"aliases":["GPT-6 Astra XHigh + SWE-2 Medium"],"api_id":null,"api_id_source_url":null,"display_name":"GPT-6 Astra XHigh + SWE-2 Medium","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"openai-cognition","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.588884131062264},"source_configuration_id":"fa2595093fa640e73d072e7d1aa7df82","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.920061132172241,"cost_usd":4.536492777821779,"refusal_rate":0.023384717860881147,"runtime_seconds":1481.89598019802,"tokens":6074134.4950494915}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.663716814159292,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.486559139784946,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.146464646464646,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":97.1880328680834},"score_percent":{"formula":"score.value * 100","value":43.22468668029613}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-5-6-luna:codex:max","model":{"aliases":["GPT-5.6 Luna"],"api_id":null,"api_id_source_url":null,"display_name":"GPT-5.6 Luna","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-5.6-luna","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4322468668029613},"source_configuration_id":"b3c56ec69185d544992fd578edb75e8c","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.971880328680834,"cost_usd":0.43802256393839384,"refusal_rate":0,"runtime_seconds":1412.260883388339,"tokens":14854524.539053895}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.722713864306785,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.540322580645161,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.373737373737374,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":94.58037017976719},"score_percent":{"formula":"score.value * 100","value":54.559127289644}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.139.0; 0.153.4"},"id":"openai:gpt-5-6-sol:codex:max","model":{"aliases":["GPT-5.6 Sol"],"api_id":null,"api_id_source_url":null,"display_name":"GPT-5.6 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-5.6-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.4,"input_per_mtok_usd":4,"output_per_mtok_usd":20,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.54559127289644},"source_configuration_id":"709f801275dedae5675b1f56e97c9df3","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9458037017976718,"cost_usd":6.353318230693071,"refusal_rate":0.000896057347670251,"runtime_seconds":1237.696445544555,"tokens":10170136.632563243}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://learn.chatgpt.com/docs/models"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.705014749262537,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.599462365591398,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.5,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":92.78388366571141},"score_percent":{"formula":"score.value * 100","value":60.14923716179784}},"effort":"high","effort_raw":"high","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-1-sol:codex:high","model":{"aliases":["GPT-6.1 Sol"],"api_id":"gpt-6.1-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6.1 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6.1-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-09T06:58:27Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6014923716179784},"source_configuration_id":"4aee8aa2d4a0bad6068644f40edae235","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9278388366571141,"cost_usd":0.8892413135295484,"refusal_rate":0.014228304550885196,"runtime_seconds":799.7652123212325,"tokens":2819911.9903514963}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://learn.chatgpt.com/docs/models"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.675516224188791,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.551075268817204,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.48989898989899,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":91.35762473536158},"score_percent":{"formula":"score.value * 100","value":57.21634943016617}},"effort":"low","effort_raw":"low","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-1-sol:codex:low","model":{"aliases":["GPT-6.1 Sol"],"api_id":"gpt-6.1-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6.1 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6.1-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-09T06:58:27Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5721634943016617},"source_configuration_id":"fa589dad83ffcdd6cce72a467e552063","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9135762473536158,"cost_usd":0.49880364510270725,"refusal_rate":0.008064516129032258,"runtime_seconds":517.8138943894386,"tokens":1713818.9181950986}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://learn.chatgpt.com/docs/models"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.696165191740413,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.577956989247312,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.53030303030303,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":92.81954641245814},"score_percent":{"formula":"score.value * 100","value":60.14750704302517}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-1-sol:codex:max","model":{"aliases":["GPT-6.1 Sol"],"api_id":"gpt-6.1-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6.1 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6.1-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-09T06:58:27Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6014750704302517},"source_configuration_id":"ea296ce785422d3b97cfff4427f2e0d2","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9281954641245814,"cost_usd":1.554119528390054,"refusal_rate":0.013440860215053764,"runtime_seconds":1463.0464224422437,"tokens":3999501.6107955067}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://learn.chatgpt.com/docs/models"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.71976401179941,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.60752688172043,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.515151515151515,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":92.39660644436648},"score_percent":{"formula":"score.value * 100","value":61.41474695571183}},"effort":"medium","effort_raw":"medium","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-1-sol:codex:medium","model":{"aliases":["GPT-6.1 Sol"],"api_id":"gpt-6.1-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6.1 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6.1-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-09T06:58:27Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6141474695571183},"source_configuration_id":"93acd59f31f8e5478686d650aa650f79","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9239660644436648,"cost_usd":0.7046746644898908,"refusal_rate":0.01075268817204301,"runtime_seconds":651.3228668866898,"tokens":2298278.745423723}},{"availability":{"access":["subscription","api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":"https://learn.chatgpt.com/docs/models"},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.731563421828909,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.610215053763441,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.545454545454545,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.27572280083257},"score_percent":{"formula":"score.value * 100","value":62.90776736822983}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-1-sol:codex:xhigh","model":{"aliases":["GPT-6.1 Sol"],"api_id":"gpt-6.1-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6.1 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6.1-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-09T06:58:27Z","cache_read_per_mtok_usd":0.1,"input_per_mtok_usd":2,"output_per_mtok_usd":10,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6290776736822983},"source_configuration_id":"ed8c1d0b0a2fe4560bbfa0a9f47e8941","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9327572280083256,"cost_usd":1.0397081014409635,"refusal_rate":0.01164874551971326,"runtime_seconds":931.3823091309132,"tokens":3212528.1094519314}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.675516224188791,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.618279569892473,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.555555555555555,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.35591564904551},"score_percent":{"formula":"score.value * 100","value":61.64504498789397}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.151.0; 0.153.4"},"id":"openai:gpt-6-astra:codex:max","model":{"aliases":["GPT-6 Astra"],"api_id":"gpt-6-astra","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6 Astra","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6-astra","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":1,"input_per_mtok_usd":10,"output_per_mtok_usd":50,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.6164504498789397},"source_configuration_id":"9e810ff3c6cce5ebaaff7c744c687a47","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9335591564904552,"cost_usd":7.46709735138597,"refusal_rate":0.01164874551971326,"runtime_seconds":1762.0330264026397,"tokens":3346611.012209417}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.63716814159292,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.443548387096774,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.151515151515152,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.06438143896165},"score_percent":{"formula":"score.value * 100","value":41.07438934016153}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-luna:codex:max","model":{"aliases":["GPT-6 Luna"],"api_id":"gpt-6-luna","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6 Luna","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6-luna","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.01,"input_per_mtok_usd":0.1,"output_per_mtok_usd":0.5,"source_url":"https://developers.openai.com/api/docs/pricing","tier":"standard_short_context"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4107438934016153},"source_configuration_id":"ac8276cd5d15ac2daacb9956ad44c132","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9606438143896164,"cost_usd":0.17591172304455457,"refusal_rate":0,"runtime_seconds":1285.9268514851478,"tokens":10218069.248074815}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.690265486725664,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.575268817204301,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.434343434343434,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":96.40119921972632},"score_percent":{"formula":"score.value * 100","value":56.662591275779974}},"effort":"max","effort_raw":"max","harness":{"name":"Codex","slug":"codex","vendor_first_party":true,"version":"0.154.0"},"id":"openai:gpt-6-sol:codex:max","model":{"aliases":["GPT-6 Sol"],"api_id":"gpt-6-sol","api_id_source_url":"https://learn.chatgpt.com/docs/models","display_name":"GPT-6 Sol","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"openai/gpt-6-sol","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"openai","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5666259127577997},"source_configuration_id":"7a0dfd2a7a9d93056f860d4675586d66","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9640119921972632,"cost_usd":2.989729004180416,"refusal_rate":0,"runtime_seconds":1338.0527051705185,"tokens":9840777.763476355}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.398230088495575,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.397849462365591,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.0353535353535354,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":0},"score_percent":{"formula":"score.value * 100","value":27.714436207156716}},"effort":"default","effort_raw":null,"harness":{"name":"Mirror","slug":"mirror","vendor_first_party":false,"version":"0.1.8-aa.1"},"id":"reflection:beam-1-0:mirror:default","model":{"aliases":["Beam 1.0"],"api_id":null,"api_id_source_url":null,"display_name":"Beam 1.0","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"reflection","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.2771443620715672},"source_configuration_id":"1b33b71cac6ac095ba7170559a03e534","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0,"cost_usd":null,"refusal_rate":0,"runtime_seconds":2541.5096302428246,"tokens":5065557.310154528}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":null},"id":"reflection:reflection-beam:opencode:default","model":{"aliases":["Reflection Beam"],"api_id":null,"api_id_source_url":null,"display_name":"Reflection Beam","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"reflection","weights":"unknown","weights_url":null},"pricing":{"as_of":"2026-10-18T07:06:28.741Z","cache_read_per_mtok_usd":null,"input_per_mtok_usd":null,"output_per_mtok_usd":null,"source_url":"https://reflection.ai/legal/beta-pricing-information","tier":"public_beta_free_with_daily_limits"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"reflection-beam-opencode-default-unspecified","source_retrieved_at":"2026-10-18T07:06:28.741Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.619469026548672,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.583333333333333,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.333333333333333,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":98.70352061053829},"score_percent":{"formula":"score.value * 100","value":51.204523107177934}},"effort":"default","effort_raw":null,"harness":{"name":"Claude Code","slug":"claude-code","vendor_first_party":false,"version":"2.1.220"},"id":"stepfun:step-5-preview:claude-code:default","model":{"aliases":["Step 5 Preview"],"api_id":null,"api_id_source_url":null,"display_name":"Step 5 Preview","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"stepfun/step-5-preview","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"stepfun","weights":"unknown","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5120452310717793},"source_configuration_id":"af5216a16f1d1e725b798b216da90479","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.987035206105383,"cost_usd":1.662215663091308,"refusal_rate":0,"runtime_seconds":3381.2690379537976,"tokens":21745804.32343233}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.648967551622419,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.583333333333333,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.176767676767677,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":90.57792630165156},"score_percent":{"formula":"score.value * 100","value":46.96895205744764}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Grok Build","slug":"grok-build","vendor_first_party":true,"version":"1.0.13"},"id":"xai:grok-4-6:grok-build:xhigh","model":{"aliases":["Grok 4.6"],"api_id":null,"api_id_source_url":null,"display_name":"Grok 4.6","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"x-ai/grok-4.6","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"xai","weights":"closed","weights_url":null},"pricing":null,"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.4696895205744764},"source_configuration_id":"70a5d300021b9986f77d4198272ef1a5","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9057792630165156,"cost_usd":3.5701432805280535,"refusal_rate":0.0026881720430107525,"runtime_seconds":1167.3387480748086,"tokens":5543624.597359739}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.725663716814159,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.629032258064516,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.333333333333333,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":93.8882976190201},"score_percent":{"formula":"score.value * 100","value":56.26764360706693}},"effort":"xhigh","effort_raw":"xhigh","harness":{"name":"Grok Build","slug":"grok-build","vendor_first_party":true,"version":"1.0.40"},"id":"xai:grok-4-7:grok-build:xhigh","model":{"aliases":["Grok 4.7"],"api_id":null,"api_id_source_url":null,"display_name":"Grok 4.7","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"x-ai/grok-4.7","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"xai","weights":"closed","weights_url":null},"pricing":{"as_of":"2026-10-10T07:01:58Z","cache_read_per_mtok_usd":0.5,"input_per_mtok_usd":2,"output_per_mtok_usd":6,"source_url":"https://docs.x.ai/developers/release-notes","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5626764360706693},"source_configuration_id":"f927d9c41b5c3021996eb14c3dd1341e","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.938882976190201,"cost_usd":8.817468730473037,"refusal_rate":0,"runtime_seconds":2354.367686468646,"tokens":14324724.283828393}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[{"evidence":"vendor-reported","n_tasks":null,"name":"DeepSWE v1.1","score":67.9,"score_scale":100,"source_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL","version":null},{"evidence":"vendor-reported","n_tasks":null,"name":"Terminal-Bench 4.0","score":28.8,"score_scale":100,"source_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL","version":null}],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":null},"id":"xiaomi:mimo-v2-6-flash:opencode:default","model":{"aliases":["MiMo-V2.6-Flash"],"api_id":null,"api_id_source_url":null,"display_name":"MiMo-V2.6-Flash","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"xiaomi/mimo-v2.6-flash","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"xiaomi","weights":"open","weights_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL"},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.0028,"input_per_mtok_usd":0.14,"output_per_mtok_usd":0.28,"source_url":"https://mimo.mi.com/models/en-US/mimo-v2.6-flash","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"mimo-v2-6-flash-opencode-default-unspecified","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[{"evidence":"vendor-reported","n_tasks":null,"name":"DeepSWE v1.1","score":71.9,"score_scale":100,"source_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL","version":null},{"evidence":"vendor-reported","n_tasks":null,"name":"Terminal-Bench 4.0","score":34.9,"score_scale":100,"source_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL","version":null}],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":null},"id":"xiaomi:mimo-v2-6-pro:opencode:default","model":{"aliases":["MiMo-V2.6-Pro"],"api_id":null,"api_id_source_url":null,"display_name":"MiMo-V2.6-Pro","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"xiaomi/mimo-v2.6-pro","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"xiaomi","weights":"open","weights_url":"https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL"},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.0036,"input_per_mtok_usd":0.435,"output_per_mtok_usd":0.87,"source_url":"https://mimo.mi.com/models/en-US/mimo-v2.6-pro","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"mimo-v2-6-pro-opencode-default-unspecified","source_retrieved_at":"2026-10-08T15:10:03Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"Not benchmarked","slug":"not-benchmarked","vendor_first_party":false,"version":null},"id":"z-ai:glm-5-3-flash:not-benchmarked:default","model":{"aliases":["GLM-5.3-Flash"],"api_id":null,"api_id_source_url":null,"display_name":"GLM-5.3-Flash","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"z-ai","weights":"unknown","weights_url":null},"pricing":{"as_of":"2026-10-11T07:03:08Z","cache_read_per_mtok_usd":0.03,"input_per_mtok_usd":0.15,"output_per_mtok_usd":0.5,"source_url":"https://docs.z.ai/guides/overview/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"glm-5.3-flash-unverified-harness-default-unspecified","source_retrieved_at":"2026-10-11T07:03:08Z","source_url":"https://docs.z.ai/guides/vlm/glm-5.3-flash","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["unknown"],"checked_at":null,"generally_available":false,"subscription_source_url":null},"comparable":false,"comparable_reason":"unmeasured","components":[],"deprecated":false,"derived":{"cache_hit_percent":null,"score_percent":null},"effort":"default","effort_raw":null,"harness":{"name":"Not benchmarked","slug":"not-benchmarked","vendor_first_party":false,"version":null},"id":"z-ai:glm-5-3-flashx:not-benchmarked:default","model":{"aliases":["GLM-5.3-FlashX"],"api_id":null,"api_id_source_url":null,"display_name":"GLM-5.3-FlashX","openrouter_checked_at":null,"openrouter_id":null,"openrouter_source_url":null,"provider":"z-ai","weights":"unknown","weights_url":null},"pricing":{"as_of":"2026-10-11T07:03:08Z","cache_read_per_mtok_usd":0.075,"input_per_mtok_usd":0.37,"output_per_mtok_usd":1.25,"source_url":"https://docs.z.ai/guides/overview/pricing","tier":"standard"},"score":{"attempts_per_task":null,"ci95":null,"n_tasks":null,"stderr":null,"value":null},"source_configuration_id":"glm-5.3-flashx-unverified-harness-default-unspecified","source_retrieved_at":"2026-10-11T07:03:08Z","source_url":"https://docs.z.ai/guides/vlm/glm-5.3-flash","status":"unmeasured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":null,"cost_usd":null,"refusal_rate":null,"runtime_seconds":null,"tokens":null}},{"availability":{"access":["api"],"checked_at":"2026-10-09T10:21:02.890728Z","generally_available":true,"subscription_source_url":null},"comparable":true,"comparable_reason":null,"components":[{"evidence":"independent","n_tasks":113,"name":"DeepSWE","score":0.613569321533923,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"1.1"},{"evidence":"independent","n_tasks":124,"name":"SWE-Atlas-QnA","score":0.594086021505376,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":null},{"evidence":"independent","n_tasks":66,"name":"Terminal-Bench","score":0.398989898989899,"score_scale":1,"source_url":"https://artificialanalysis.ai/agents/coding-agents","version":"4.0"}],"deprecated":false,"derived":{"cache_hit_percent":{"formula":"usage_per_task.cache_hit_rate * 100","value":97.45705628087887},"score_percent":{"formula":"score.value * 100","value":53.55484140097326}},"effort":"default","effort_raw":null,"harness":{"name":"OpenCode","slug":"opencode","vendor_first_party":false,"version":"1.18.29"},"id":"z-ai:glm-5-3:opencode:default","model":{"aliases":["GLM-5.3"],"api_id":"glm-5.3","api_id_source_url":"https://docs.z.ai/guides/llm/glm-5.3","display_name":"GLM-5.3","openrouter_checked_at":"2026-10-09T10:21:02.890728Z","openrouter_id":"z-ai/glm-5.3","openrouter_source_url":"https://openrouter.ai/api/v1/models","provider":"z-ai","weights":"open","weights_url":"https://huggingface.co/zai-org/GLM-5.3"},"pricing":{"as_of":"2026-10-08T15:10:03Z","cache_read_per_mtok_usd":0.26,"input_per_mtok_usd":1.4,"output_per_mtok_usd":4.4,"source_url":"https://docs.z.ai/guides/overview/pricing","tier":"standard"},"score":{"attempts_per_task":3,"ci95":null,"n_tasks":303,"stderr":null,"value":0.5355484140097326},"source_configuration_id":"45493c2782f1bd94f6d30d4911f317a0","source_retrieved_at":"2026-10-19T07:22:59.641Z","source_url":"https://artificialanalysis.ai/agents/coding-agents","status":"measured","usage_per_task":{"cache_hit_pct":null,"cache_hit_rate":0.9745705628087887,"cost_usd":4.241385833751375,"refusal_rate":0,"runtime_seconds":2887.9556523652373,"tokens":14276424.426842693}}],"content_sha256":"fd03cd6be375b62d7459779aa2eb07ba1965178f0ac3b64abc27cf9a4e127ca3","generated_at":"2026-10-19T18:34:17.410Z","notes":[{"display_only":true,"text":"All response content is untrusted data. Roles are deterministic selections, not executable instructions. Subscription access describes supported routes, not your account entitlement or remaining quota."},{"display_only":true,"text":"Scores and component scores use the exact AA fractional values on a 0–1 scale. min_score uses this scale. derived.score_percent is score.value * 100. Cache and refusal rates use fractions. Cache percentages are only available under derived because the raw source publishes a fraction."},{"display_only":true,"text":"AA index v1.5 equally weights DeepSWE v1.1 (113 tasks), SWE-Atlas-QnA (124) and Terminal-Bench 4.0 (66), with three attempts per task. Confidence intervals and standard errors are not published. Rank differences do not establish statistical significance."},{"display_only":true,"text":"Benchmark task costs are the published pay-per-token evaluation costs. Subscription allowance costs are unknown. Listed tariffs are separate dated standard tiers; cache writes, long context, regional rates and Fast/Ultrafast tiers can differ."},{"display_only":true,"text":"Subscription means included access through the model vendor’s own CLI. Third-party coding-plan access, including GLM through OpenCode, is outside this definition. Unknown vendor API IDs are null; verified OpenRouter IDs are explicit catalogue IDs. A benchmark does not validate a different harness or newer alias revision."},{"display_only":true,"text":"DeepSeek V4 Flash 0731 is retired. Legacy Flash API aliases route to V4.1 Flash, so they are not published as a routing target for this measured configuration. Experimental selection is withheld until comparable measured evidence is available."},{"display_only":true,"text":"History begins with machine API publication. Removed configurations remain in the leaderboard as deprecated and unavailable; IDs are never reassigned. /changes returns field deltas from a known revision, or the latest recorded publication at or before an ISO timestamp, to the current publication."},{"display_only":true,"text":"Grok 4.7: Standard public xAI API tariff in USD per 1M tokens below 200k prompt tokens. Above 200k: $4 input, $1 cached input, $12 output. Grok 4.7 Fast is a distinct processing tier available only in Cursor and Grok Build, not the public xAI API. AA's Grok Build benchmark cost is unchanged; no tariff-based task-cost recalculation."},{"display_only":true,"text":"Gemini 4 Argon: Gemini Developer API Standard paid-tier introductory USD rates per 1M tokens through 2026-11-30. From 2026-12-01: $4 input, $0.20 cached input, $20 output including thinking. Unified 1M context; implicit caching only, no explicit caching. Batch, Flex, Priority and invite-only Contributor tiers are distinct. Not on the Free tier or Google AI Pro plan. No region-specific adjustment inferred. The AA Antigravity CLI task cost remains the published evaluation snapshot."},{"display_only":true,"text":"Sonnet 5.5: Current cache read tariff differs from the launch page ($0.20). Benchmark cost is preserved as published."},{"display_only":true,"text":"GPT-6 Astra: Standard short-context tariff. Long context: $20 input, $2 cache, $75 output. Cache writes separate."},{"display_only":true,"text":"Fable 5.1: Standard first-party API. Fallback usage can change effective cost."},{"display_only":true,"text":"Opus 5.5: Standard first-party USD tariff per 1M tokens; cache writes cost extra. Current documentation lists Opus 5.5 fast mode at $8 input/$40 output, distinct from this $4/$20 standard tariff. It is now documented on first-party Claude API and Azure-hosted Claude in Microsoft Foundry, not Claude Platform on AWS. No fast-mode benchmark cost or runtime is inferred."},{"display_only":true,"text":"GPT-6.1 Sol: Standard short-context tariff. Long context: $4 input, $0.20 cache, $15 output. Cache writes separate. Optional Ultrafast service tier for GPT-6.1 Sol announced 8 October 2026: short context per 1M tokens $12 input, $0.60 cached input, $15 cache writes, $60 output; long context $24/$1.20/$30/$90 respectively. Six times the standard token tariff; distinct processing tier, not a new model or a rerun of the recorded AA benchmark. Global processing and US/EU data residency supported; regional processing surcharge applies separately."},{"display_only":true,"text":"GPT-6 Luna: Standard short-context tariff. Long context costs more."},{"display_only":true,"text":"Haiku 5.5: Prompts up to 100,000 tokens. Longer prompts: $0.50 input, $0.05 cache, $2.50 output."},{"display_only":true,"text":"Qwen3.8 Max: Singapore region, implicit cache. Snapshot 0902 listed at these rates; exact AA snapshot not inferred. Frankfurt/global rates differ."},{"display_only":true,"text":"GPT-5.6 Sol: Short context. Promotional price listed through at least 21 November 2026. Long context costs more."},{"display_only":true,"text":"Opus 5: Standard first-party API; cache writes cost extra."},{"display_only":true,"text":"Kimi K3: First-party published API tariff; no subscription equivalence assumed."},{"display_only":true,"text":"GLM-5.3: Standard tariff; custom GLM-5.3 weight licence. Cache-storage promotion has separate terms."},{"display_only":true,"text":"MiMo-V2.6-Pro: Standard USD tariff. Released Pro-RL checkpoint is MIT; hosted API revision equivalence is not established."},{"display_only":true,"text":"MiMo-V2.6-Flash: Standard USD tariff. Released Flash-RL checkpoint is MIT; API revision equivalence is not established."},{"display_only":true,"text":"Mistral Large 4: Current USD preview sale per 1M tokens: $0.68 input, $0.07 cached input, $2.09 output. Listed non-sale rates are $1.36/$0.14/$4.18. Sale expiry and region-specific adjustments remain unverified. The documentation labels the model open-weight, but a released checkpoint artifact was not established in this bounded audit; weights status remains unknown."},{"display_only":true,"text":"Reflection Beam: Reflection's public beta page still states free access subject to daily usage limits; paid production token rates remain unknown, not zero. The publisher now links a Beam weights collection. Exact checkpoint artifacts and licence were not fully checked within the review bound; no continuing 'weights pending' claim is made."},{"display_only":true,"text":"GLM-5.3-Flash: Published USD API rates per 1M tokens. Cached-input storage temporarily free; expiry and regional qualifications unknown. Official documentation confirms GLM Coding Plan availability."},{"display_only":true,"text":"GLM-5.3-FlashX: Published USD API rates per 1M tokens. Cached-input storage temporarily free; expiry and regional qualifications unknown. Official documentation states FlashX is not yet included in the GLM Coding Plan."},{"display_only":true,"text":"Muse Wonder: Current Meta Model API tariff in USD per 1M tokens for muse-wonder-1.0. One-million-token context. Available in Meta Model API and Muse Code; no regional or promotional adjustment inferred."}],"revision":"agent-v1-eaa0b33df86d5fe87842668049fc5f1ba971502b6a83158081df9fae96898169","roles":[{"confidence":"provisional","configuration_id":"openai:gpt-6-1-sol:codex:xhigh","role":"default_worker","rule":"eligible := measured AND comparable AND available AND NOT deprecated; argmax(score.value WHERE score.value > 0.5 AND cost_usd <= 2); ties := cost_usd ASC, id ASC"},{"confidence":"measured","configuration_id":"anthropic:sonnet-5-5:claude-code:max","role":"max_capability","rule":"eligible := measured AND comparable AND available AND NOT deprecated; argmax(score.value); ties := cost_usd ASC NULLS LAST, id ASC"},{"confidence":"provisional","configuration_id":"openai:gpt-6-1-sol:codex:low","role":"bulk_worker","rule":"eligible := measured AND comparable AND available AND NOT deprecated; argmin(cost_usd WHERE score.value > 0.5 AND cost_usd IS NOT NULL); ties := score.value DESC, id ASC"},{"confidence":"measured","configuration_id":"z-ai:glm-5-3:opencode:default","role":"open_weight","rule":"eligible := measured AND comparable AND available AND NOT deprecated; argmax(score.value WHERE weights = open); ties := cost_usd ASC NULLS LAST, id ASC"},{"confidence":"withheld","configuration_id":null,"role":"experimental","rule":"NULL"}],"schema_version":"1.0","signature":{"algorithm":"Ed25519","encoding":"base64","key_id":"b354c52cb8f656ee8b1d26e591f52c441ac206943e7a141b1e63a676d6dd3070","public_key_url":"/.well-known/leaderboard-key.pub","signed":"content_sha256_bytes","value":"CmQbjT7U6wTcUEwTeTU5vI1OkiHy5nflV8XkRh9rbtwEu163VOpAAlsE80UFBO+njijS9c0ijcpEIK9gaGOjBw=="},"source":{"components":[{"n_tasks":113,"name":"DeepSWE","version":"1.1"},{"n_tasks":124,"name":"SWE-Atlas-QnA","version":null},{"n_tasks":66,"name":"Terminal-Bench","version":"4.0"}],"index_version":"1.5","methodology_url":"https://artificialanalysis.ai/methodology/coding-agents-benchmarking","name":"Artificial Analysis","rate_scale":1,"retrieved_at":"2026-10-19T07:22:59.641Z","score_scale":1,"url":"https://artificialanalysis.ai/agents/coding-agents"},"stale":true,"stale_reason":"aa_snapshot_older_than_48_hours"}