Compare models
Comparing 2models. Drop the URL into a doc — it's permalinked.
| Field | deepseek/deepseek-reasoner | deepseek/deepseek-chat |
|---|---|---|
| Provider | deepseek | deepseek |
| Model ID | deepseek-reasoner | deepseek-chat |
| Context | 64K | 164K |
| Max output | 8K | 8K |
| Input / 1M | $0.55 | $0.32 |
| Output / 1M | $2.19 | $0.89 |
| Cached input / 1M | $0.14 | $0.07 |
| Avg cost / 1M | $1.37 | $0.60 |
| Speed | 30 t/s | 50 t/s |
| Quality index | 76.0 | 72.0 |
| MMLU | 90.8 | 87.1 |
| GPQA | 71.5 | 59.1 |
| HumanEval | 86.4 | 89.0 |
| MATH | 97.3 | 90.2 |
| SWE-bench | — | — |
| Arena Elo | — | — |
| Tools | — | ✓ |
| Vision | — | — |
| Thinking | ✓ | — |
| Streaming | ✓ | ✓ |
| JSON mode | — | ✓ |
| Structured output | — | ✓ |
| Prompt cache | — | — |
Same data, in your terminal: relay models compare r1 deepseek-v3