← All comparisons
GPT-5 (medium) vs Claude 3.5 Sonnet (Oct '24)
OpenAI vs Anthropic — side-by-side benchmark comparison
| GPT-5 (medium) | Claude 3.5 Sonnet (Oct '24) | |
|---|---|---|
| Intelligence Index | 42.0 | 15.9 |
| Coding Index | 38.9 | 30.2 |
| Math Index | 91.7 | — |
| Output speed (tok/s) | 86.4 | 0.0 |
| Blended price ($/1M) | $3.44 | $6.56 |
| Time to first token (s) | 37.15s | 0.00s |
| aime | 91.7% | 15.7% |
| aime 25 | 91.7% | — |
| artificial analysis coding index | 38.90 | 30.20 |
| artificial analysis intelligence index | 42.00 | 15.90 |
| artificial analysis math index | 91.70 | — |
| gpqa | 84.2% | 59.9% |
| hle | 23.5% | 3.9% |
| ifbench | 70.6% | — |
| lcr | 72.8% | — |
| livecodebench | 70.3% | 38.1% |
| math 500 | 99.1% | 77.1% |
| mmlu pro | 86.7% | 77.2% |
| scicode | 41.1% | 36.6% |
| tau2 | 86.5% | — |
| terminalbench hard | 37.9% | — |
Benchmark data from Artificial Analysis.