← All comparisons
Gemini 2.5 Flash (Reasoning) vs Claude 3.5 Sonnet (Oct '24)
Google vs Anthropic — side-by-side benchmark comparison
| Gemini 2.5 Flash (Reasoning) | Claude 3.5 Sonnet (Oct '24) | |
|---|---|---|
| Intelligence Index | 27.0 | 15.9 |
| Coding Index | 22.2 | 30.2 |
| Math Index | 73.3 | — |
| Output speed (tok/s) | 205.5 | 0.0 |
| Blended price ($/1M) | $0.85 | $6.56 |
| Time to first token (s) | 10.67s | 0.00s |
| aime | 82.3% | 15.7% |
| aime 25 | 73.3% | — |
| artificial analysis coding index | 22.20 | 30.20 |
| artificial analysis intelligence index | 27.00 | 15.90 |
| artificial analysis math index | 73.30 | — |
| gpqa | 79.0% | 59.9% |
| hle | 11.1% | 3.9% |
| ifbench | 50.3% | — |
| lcr | 61.7% | — |
| livecodebench | 69.5% | 38.1% |
| math 500 | 98.1% | 77.1% |
| mmlu pro | 83.2% | 77.2% |
| scicode | 39.4% | 36.6% |
| tau2 | 31.6% | — |
| terminalbench hard | 13.6% | — |
Benchmark data from Artificial Analysis.