TPSTokens Per Second

Claude Sonnet 4.5

Anthropic🇺🇸 United States
active

Balanced intelligence and speed. Recommended for most tasks.

Compare with other models →
Context window200K tokens
Input / 1M tokens$3
Output / 1M tokens$15

Version History

claude-sonnet-4-5-20260121patch

Claude Sonnet 4.5 January patch with stability improvements for long agentic sessions and better tool use across multi-turn workflows.

claude-sonnet-4-5major

Claude Sonnet 4.5 with improved agentic performance and hybrid reasoning.

Benchmark Scores

Full leaderboard →
78.0%
AIME 2024
87.0%
AIME 2025
93.5%
DocVQA
83.4%
GPQA
90.5%
HumanEval
69.1%
LiveCodeBench
88.3%
MATH
89.1%
MMLU
73.8%
MMLU-Pro
77.8%
MMMU
82.0 tokens_per_sec
Speed (tok/s)
77.2%
SWE-bench Verified