← All comparisons
Gemini 2.5 Flash (Reasoning) vs Claude 4.5 Sonnet (Reasoning)
Google vs Anthropic — side-by-side benchmark comparison
| Gemini 2.5 Flash (Reasoning) | Claude 4.5 Sonnet (Reasoning) | |
|---|---|---|
| Intelligence Index | 27.0 | 43.0 |
| Coding Index | 22.2 | 38.6 |
| Math Index | 73.3 | 88.0 |
| Output speed (tok/s) | 205.5 | 55.0 |
| Blended price ($/1M) | $0.85 | $6.56 |
| Time to first token (s) | 10.67s | 7.02s |
| aime | 82.3% | — |
| aime 25 | 73.3% | 88.0% |
| artificial analysis coding index | 22.20 | 38.60 |
| artificial analysis intelligence index | 27.00 | 43.00 |
| artificial analysis math index | 73.30 | 88.00 |
| gpqa | 79.0% | 83.4% |
| hle | 11.1% | 17.3% |
| ifbench | 50.3% | 57.3% |
| lcr | 61.7% | 65.7% |
| livecodebench | 69.5% | 71.4% |
| math 500 | 98.1% | — |
| mmlu pro | 83.2% | 87.5% |
| scicode | 39.4% | 44.7% |
| tau2 | 31.6% | 78.1% |
| terminalbench hard | 13.6% | 35.6% |
Benchmark data from Artificial Analysis.