Compare models
Comparing 2models. Drop the URL into a doc — it's permalinked.
| Field | anthropic/claude-sonnet-4-5 | deepseek/deepseek-chat |
|---|---|---|
| Provider | anthropic | deepseek |
| Model ID | claude-sonnet-4-5 | deepseek-chat |
| Context | 200K | 164K |
| Max output | 64K | 8K |
| Input / 1M | $3.00 | $0.32 |
| Output / 1M | $15.00 | $0.89 |
| Cached input / 1M | $0.30 | $0.07 |
| Avg cost / 1M | $9.00 | $0.60 |
| Speed | 95 t/s | 50 t/s |
| Quality index | 73.0 | 72.0 |
| MMLU | 88.7 | 87.1 |
| GPQA | 71.5 | 59.1 |
| HumanEval | 89.1 | 89.0 |
| MATH | 91.8 | 90.2 |
| SWE-bench | 49.0 | — |
| Arena Elo | — | — |
| Tools | ✓ | ✓ |
| Vision | ✓ | — |
| Thinking | ✓ | — |
| Streaming | ✓ | ✓ |
| JSON mode | — | ✓ |
| Structured output | ✓ | ✓ |
| Prompt cache | ✓ | — |
Same data, in your terminal: relay models compare sonnet-4.5 deepseek-v3