Claude Opus 4

Anthropic🇺🇸 United States
active

Anthropic's most capable Claude 4 model.

Context window200K tokens
Input / 1M tokens$15
Output / 1M tokens$75

Version History

claude-opus-4-0major

Major new generation. Strongest reasoning and coding of any Anthropic model.

Benchmark Scores

Full leaderboard →
93.0%
DocVQA
86.5%
GPQA
92.4%
HumanEval
70.0%
LiveCodeBench
89.5%
MMLU
74.0%
MMMU
40.0 tokens_per_sec
Speed (tok/s)
72.5%
SWE-bench Verified