← All comparisons
Claude 4.5 Haiku (Non-reasoning) vs GPT-4.1
Anthropic vs OpenAI — side-by-side benchmark comparison
| Claude 4.5 Haiku (Non-reasoning) | GPT-4.1 | |
|---|---|---|
| Intelligence Index | 31.0 | 26.3 |
| Coding Index | 29.6 | 21.8 |
| Math Index | 39.0 | 34.7 |
| Output speed (tok/s) | 105.8 | 137.8 |
| Blended price ($/1M) | $2.19 | $3.50 |
| Time to first token (s) | 0.65s | 0.58s |
| aime | — | 43.7% |
| aime 25 | 39.0% | 34.7% |
| artificial analysis coding index | 29.60 | 21.80 |
| artificial analysis intelligence index | 31.00 | 26.30 |
| artificial analysis math index | 39.00 | 34.70 |
| gpqa | 64.6% | 66.6% |
| hle | 4.3% | 4.6% |
| ifbench | 42.0% | 43.0% |
| lcr | 43.7% | 61.0% |
| livecodebench | 51.1% | 45.7% |
| math 500 | — | 91.3% |
| mmlu pro | 80.0% | 80.6% |
| scicode | 34.4% | 38.1% |
| tau2 | 32.5% | 47.1% |
| terminalbench hard | 27.3% | 13.6% |
Benchmark data from Artificial Analysis.