← All comparisons
Llama 3.1 Nemotron Instruct 70B vs Granite 4.0 H 1B
NVIDIA vs IBM — side-by-side benchmark comparison
| Llama 3.1 Nemotron Instruct 70B | Granite 4.0 H 1B | |
|---|---|---|
| Intelligence Index | 13.4 | 8.0 |
| Coding Index | 10.8 | 2.7 |
| Math Index | 11.0 | 6.3 |
| Output speed (tok/s) | 292.9 | 0.0 |
| Blended price ($/1M) | $1.20 | $0.00 |
| Time to first token (s) | 0.24s | 0.00s |
| aime | 24.7% | — |
| aime 25 | 11.0% | 6.3% |
| artificial analysis coding index | 10.80 | 2.70 |
| artificial analysis intelligence index | 13.40 | 8.00 |
| artificial analysis math index | 11.00 | 6.30 |
| gpqa | 46.5% | 26.3% |
| hle | 4.6% | 5.0% |
| ifbench | 30.7% | 26.2% |
| lcr | 7.0% | 6.3% |
| livecodebench | 16.9% | 11.5% |
| math 500 | 73.3% | — |
| mmlu pro | 69.0% | 27.7% |
| scicode | 23.3% | 8.2% |
| tau2 | 23.1% | 19.6% |
| terminalbench hard | 4.5% | 0.0% |
Benchmark data from Artificial Analysis.