← All comparisons
Gemini 3.1 Flash-Lite vs Hermes 4 - Llama-3.1 70B (Non-reasoning)
Google vs Nous Research — side-by-side benchmark comparison
| Gemini 3.1 Flash-Lite | Hermes 4 - Llama-3.1 70B (Non-reasoning) | |
|---|---|---|
| Intelligence Index | 33.5 | 12.6 |
| Coding Index | 30.1 | 9.2 |
| Math Index | — | 11.3 |
| Output speed (tok/s) | 304.9 | 94.3 |
| Blended price ($/1M) | $0.56 | $0.20 |
| Time to first token (s) | 5.00s | 0.61s |
| aime | — | — |
| aime 25 | — | 11.3% |
| artificial analysis coding index | 30.10 | 9.20 |
| artificial analysis intelligence index | 33.50 | 12.60 |
| artificial analysis math index | — | 11.30 |
| gpqa | 82.2% | 49.1% |
| hle | 16.2% | 3.6% |
| ifbench | 77.2% | 29.0% |
| lcr | 65.3% | 2.0% |
| livecodebench | — | 26.9% |
| math 500 | — | — |
| mmlu pro | — | 66.4% |
| scicode | 41.9% | 27.7% |
| tau2 | 31.3% | 21.6% |
| terminalbench hard | 24.2% | 0.0% |
Benchmark data from Artificial Analysis.