← All comparisons
GPT-5 Codex (high) vs Claude 4.1 Opus (Non-reasoning)
OpenAI vs Anthropic — side-by-side benchmark comparison
| GPT-5 Codex (high) | Claude 4.1 Opus (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 44.6 | 36.0 |
| Coding Index | 38.9 | — |
| Math Index | 98.7 | — |
| Output speed (tok/s) | 173.2 | 44.7 |
| Blended price ($/1M) | $3.44 | $32.81 |
| Time to first token (s) | 7.64s | 1.63s |
| aime | — | — |
| aime 25 | 98.7% | — |
| artificial analysis coding index | 38.90 | — |
| artificial analysis intelligence index | 44.60 | 36.00 |
| artificial analysis math index | 98.70 | — |
| gpqa | 83.7% | — |
| hle | 25.6% | — |
| ifbench | 74.1% | — |
| lcr | 69.0% | — |
| livecodebench | 84.0% | — |
| math 500 | — | — |
| mmlu pro | 86.5% | — |
| scicode | 40.9% | — |
| tau2 | 86.8% | — |
| terminalbench hard | 37.9% | — |
Benchmark data from Artificial Analysis.