Home Model series Claude Sonnet Claude Sonnet 4.6 Claude Sonnet 4.6 · DeepSearchQA · 900 prompts;单代理与多代理按协议区分 · Scores and configurations Claude Sonnet 4.6 · DeepSearchQA · 900 prompts;单代理与多代理按协议区分 · Scores and configurations Claude Sonnet 4.6 record for DeepSearchQA · 900 prompts;单代理与多代理按协议区分. 2 score records for this release, including 2 exact values (may include historical records). Includes 2 source locations.
Scores and evaluation conditions Claude Sonnet 4.6 · DeepSearchQA scores and configurations Metric and score Configuration and evaluation conditions Sources F1 score 89.2% Reported 89.2% · Unit % Published Unconfirmed
Adaptive thinking, max effort Settings Adaptive thinking enabled; max effort; default temperature and top_p unless an evaluation specifies otherwise. Unknowns Context window is evaluation-dependent; Table 2.1.A states it does not exceed 1M tokens. Tools Tools allowed Reasoning budget max Framework not recorded Protocol DeepSearchQA, 900 prompts; single-agent configuration. Conditions Web search, web fetch, programmatic tool calling, context compaction at 50K tokens up to 10M total tokens, adaptive thinking and max effort. View this score in rankings F1 score 91.1% Reported 91.1% · Unit % Published Unconfirmed
Adaptive thinking, max effort Settings Adaptive thinking enabled; max effort; default temperature and top_p unless an evaluation specifies otherwise. Unknowns Context window is evaluation-dependent; Table 2.1.A states it does not exceed 1M tokens. Tools Tools allowed Reasoning budget max Framework not recorded Protocol DeepSearchQA, 900 prompts; multi-agent configuration. Conditions F1 score; orchestrator and subagents use programmatic tool calling, 50K compaction and up to 3M tokens each; max reasoning effort.
Benchmark versions and metrics Aliases DeepSearchQA、DeepSearchQA with multi-agents
F1 score · % · Higher is better Accuracy · % · Higher is better