Home Model series Claude Haiku Claude Haiku 4.5 Claude Haiku 4.5 · τ²-bench · Retail section · Scores and configurations Claude Haiku 4.5 · τ²-bench · Retail section · Scores and configurations Claude Haiku 4.5 record for τ²-bench · Retail section. 1 score record for this release, including 1 exact value (may include historical records). Includes 1 source location.
Scores and evaluation conditions Claude Haiku 4.5 · τ²-bench scores and configurations Metric and score Configuration and evaluation conditions Sources Success rate 83.2% Reported 83.2% · Unit % Published Unconfirmed
发布表报告值 Settings SWE-bench: full 500 problems, 50 trials, no test-time compute, 128K thinking budget, default sampling; Terminal-Bench: Terminus 2, 11 runs split between no thinking and 32K thinking; OSWorld: OSWorld-Verified, 100 steps, 4 runs; GPQA/MMMLU/MMMU follow source table. Unknowns AIME 的 Python 工具使用情况按两个分列配置保存;其余各项的逐行配置以主文脚注为准。 Tools Tools allowed Reasoning budget Thinking budget 128K Framework not recorded Protocol τ²-bench Retail。 Conditions 按发布表及模型系统卡中 benchmark 说明;航空策略提示有修订,见报告。 View this score in rankings
Benchmark versions and metrics Aliases tau2-bench Retail、τ²-bench Retail、τ2-bench Retail、tau2-bench
Success rate · % · Higher is better