Home Model series Claude Sonnet Claude Sonnet 4.6 Claude Sonnet 4.6 · ARC-AGI-1 · ARC Prize Foundation private dataset · Scores and configurations Claude Sonnet 4.6 · ARC-AGI-1 · ARC Prize Foundation private dataset · Scores and configurations Claude Sonnet 4.6 record for ARC-AGI-1 · ARC Prize Foundation private dataset. 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 · ARC-AGI-1 scores and configurations Metric and score Configuration and evaluation conditions Sources Accuracy 86.5% Reported 86.50% · Unit % Published Unconfirmed
High effort, 120K thinking budget Settings High effort with 120K thinking tokens; ARC Prize Foundation private dataset evaluation. Unknowns none recorded Tools Tool access not recorded Reasoning budget high · Thinking budget 120K Framework not recorded Protocol ARC Prize Foundation private dataset. Conditions High effort; 120K thinking tokens. Figure caption rounds the value to 86.5%. Accuracy 86% Reported 86.0% · Unit % Published Unconfirmed
ARC · Max effort Settings ARC 图表与同表脚注明确 Max effort。 Unknowns 未说明 thinking mode 与 token budget。 Tools Tool access not recorded Reasoning budget max Framework not recorded Protocol ARC Prize Foundation private dataset. Conditions Figure 2.7.A labels this comparison point Claude Sonnet 4.6 (Max). Adaptive thinking and thinking-token budget are not stated for this figure result. View this score in rankings
Benchmark versions and metrics Aliases ARC-AGI-1
Accuracy · % · Higher is better