Claude Sonnet 4
FrontierMultimodalAPI Available
Benchmark Scores
89.4/100MMLU89.5
HumanEval93.9
Math80.1
Reasoning89.8
Coding93.5
About Claude Sonnet 4
Claude Sonnet 4 is Anthropic's best balance of performance and cost, significantly improving on Claude 3.5 Sonnet with better reasoning, reduced sycophancy, and improved instruction following across all domains.
Strengths
- +Excellent coding
- +More predictable and steerable
- +Strong analytical reasoning
- +Great value at its price
Weaknesses
- -Not as powerful as Opus 4 for hardest tasks
- -Occasionally over-hedges
Pricing
Per 1M tokens
Input$3.00
Output$15.00
Quick Facts
- Context Window
- 200K
- Parameters
- Undisclosed
- Release Date
- 2025-05-22
- Category
- Frontier
Compare Models
See how Claude Sonnet 4 stacks up against other models