Claude 3.7 Sonnet vs Qwen 2.5 Max
A direct, empirical evaluation of Claude 3.7 Sonnet (Anthropic) against Qwen 2.5 Max (Alibaba Cloud).
Claude 3.7 Sonnet
Developed by Anthropic · Proprietary parameters
Enterprise software engineering, full-repo refactoring, and agentic workflows.
Qwen 2.5 Max
Developed by Alibaba Cloud · Proprietary (~700B) parameters
Cross-lingual document extraction and structured structured output generation.
| Metric | Claude 3.7 Sonnet | Qwen 2.5 Max |
|---|---|---|
| SWE-bench Verified | 70.3% | 51.4% |
| MMLU-Pro | 82.1% | 77.2% |
| MATH-500 | 96.2% | 89.4% |
| Context Window | 200,000 tokens | 128,000 tokens |
| Input Pricing (per 1M) | $3.00 | $1.60 |
| Output Pricing (per 1M) | $15.00 | $6.40 |