← All comparisons
Granite 4.0 H 1B vs Hermes 3 - Llama-3.1 70B
IBM vs Nous Research — side-by-side benchmark comparison
| Granite 4.0 H 1B | Hermes 3 - Llama-3.1 70B | |
|---|---|---|
| Intelligence Index | 8.0 | 10.6 |
| Coding Index | 2.7 | — |
| Math Index | 6.3 | — |
| Output speed (tok/s) | 0.0 | 33.2 |
| Blended price ($/1M) | $0.00 | $0.30 |
| Time to first token (s) | 0.00s | 0.38s |
| aime | — | 2.3% |
| aime 25 | 6.3% | — |
| artificial analysis coding index | 2.70 | — |
| artificial analysis intelligence index | 8.00 | 10.60 |
| artificial analysis math index | 6.30 | — |
| gpqa | 26.3% | 40.1% |
| hle | 5.0% | 4.1% |
| ifbench | 26.2% | — |
| lcr | 6.3% | — |
| livecodebench | 11.5% | 18.8% |
| math 500 | — | 53.8% |
| mmlu pro | 27.7% | 57.1% |
| scicode | 8.2% | 23.1% |
| tau2 | 19.6% | — |
| terminalbench hard | 0.0% | — |
Benchmark data from Artificial Analysis.