Compare models
Comparing 2models. Drop the URL into a doc — it's permalinked.
| Field | deepseek/deepseek-reasoner | anthropic/claude-sonnet-4-5 |
|---|---|---|
| Provider | deepseek | anthropic |
| Model ID | deepseek-reasoner | claude-sonnet-4-5 |
| Context | 64K | 200K |
| Max output | 8K | 64K |
| Input / 1M | $0.55 | $3.00 |
| Output / 1M | $2.19 | $15.00 |
| Cached input / 1M | $0.14 | $0.30 |
| Avg cost / 1M | $1.37 | $9.00 |
| Speed | 30 t/s | 95 t/s |
| Quality index | 76.0 | 73.0 |
| MMLU | 90.8 | 88.7 |
| GPQA | 71.5 | 71.5 |
| HumanEval | 86.4 | 89.1 |
| MATH | 97.3 | 91.8 |
| SWE-bench | — | 49.0 |
| Arena Elo | — | — |
| Tools | — | ✓ |
| Vision | — | ✓ |
| Thinking | ✓ | ✓ |
| Streaming | ✓ | ✓ |
| JSON mode | — | — |
| Structured output | — | ✓ |
| Prompt cache | — | ✓ |
Same data, in your terminal: relay models compare r1 sonnet-4.5