← All comparisons
Gemini 3 Pro Preview (high) vs Claude 4 Opus (Non-reasoning)
Google vs Anthropic — side-by-side benchmark comparison
| Gemini 3 Pro Preview (high) | Claude 4 Opus (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 48.4 | 33.0 |
| Coding Index | 46.5 | — |
| Math Index | 95.7 | 36.3 |
| Output speed (tok/s) | 0.0 | 43.3 |
| Blended price ($/1M) | $4.50 | $32.81 |
| Time to first token (s) | 0.00s | 1.66s |
| aime | — | 56.3% |
| aime 25 | 95.7% | 36.3% |
| artificial analysis coding index | 46.50 | — |
| artificial analysis intelligence index | 48.40 | 33.00 |
| artificial analysis math index | 95.70 | 36.30 |
| gpqa | 90.8% | 70.1% |
| hle | 37.2% | 5.9% |
| ifbench | 70.4% | 43.3% |
| lcr | 70.7% | 36.0% |
| livecodebench | 91.7% | 54.2% |
| math 500 | — | 94.1% |
| mmlu pro | 89.8% | 86.0% |
| scicode | 56.1% | 40.9% |
| tau2 | 87.1% | — |
| terminalbench hard | 41.7% | — |
Benchmark data from Artificial Analysis.