← All comparisons
GPT-5.2 Codex (xhigh) vs Claude 4.1 Opus (Reasoning)
OpenAI vs Anthropic — side-by-side benchmark comparison
| GPT-5.2 Codex (xhigh) | Claude 4.1 Opus (Reasoning) | |
|---|---|---|
| Intelligence Index | 49.0 | 42.0 |
| Coding Index | 43.0 | 36.5 |
| Math Index | — | 80.3 |
| Output speed (tok/s) | 111.3 | 44.5 |
| Blended price ($/1M) | $4.81 | $32.81 |
| Time to first token (s) | 1.33s | 8.55s |
| aime | — | — |
| aime 25 | — | 80.3% |
| artificial analysis coding index | 43.00 | 36.50 |
| artificial analysis intelligence index | 49.00 | 42.00 |
| artificial analysis math index | — | 80.30 |
| gpqa | 89.9% | 80.9% |
| hle | 33.5% | 11.9% |
| ifbench | 77.6% | 55.4% |
| lcr | 75.7% | 66.3% |
| livecodebench | — | 65.4% |
| math 500 | — | — |
| mmlu pro | — | 88.0% |
| scicode | 54.6% | 40.9% |
| tau2 | 92.1% | 71.4% |
| terminalbench hard | 37.1% | 34.3% |
Benchmark data from Artificial Analysis.