← All comparisons
Claude 4.1 Opus (Non-reasoning) vs DeepSeek V3.1 Terminus (Reasoning)
Anthropic vs DeepSeek — side-by-side benchmark comparison
| Claude 4.1 Opus (Non-reasoning) | DeepSeek V3.1 Terminus (Reasoning) | |
|---|---|---|
| Intelligence Index | 36.0 | 33.9 |
| Coding Index | — | 33.7 |
| Math Index | — | 89.7 |
| Output speed (tok/s) | 44.7 | 0.0 |
| Blended price ($/1M) | $32.81 | $1.91 |
| Time to first token (s) | 1.63s | 0.00s |
| aime | — | — |
| aime 25 | — | 89.7% |
| artificial analysis coding index | — | 33.70 |
| artificial analysis intelligence index | 36.00 | 33.90 |
| artificial analysis math index | — | 89.70 |
| gpqa | — | 79.2% |
| hle | — | 15.2% |
| ifbench | — | 57.0% |
| lcr | — | 65.0% |
| livecodebench | — | 79.8% |
| math 500 | — | — |
| mmlu pro | — | 85.1% |
| scicode | — | 40.6% |
| tau2 | — | 37.1% |
| terminalbench hard | — | 30.3% |
Benchmark data from Artificial Analysis.