← All comparisons
Granite 4.1 8B vs DeepSeek R1 Distill Qwen 32B
IBM vs DeepSeek — side-by-side benchmark comparison
| Granite 4.1 8B | DeepSeek R1 Distill Qwen 32B | |
|---|---|---|
| Intelligence Index | 12.4 | 17.2 |
| Coding Index | 7.3 | — |
| Math Index | — | 63.0 |
| Output speed (tok/s) | 111.5 | 0.0 |
| Blended price ($/1M) | $0.06 | $0.00 |
| Time to first token (s) | 0.41s | 0.00s |
| aime | — | 68.7% |
| aime 25 | — | 63.0% |
| artificial analysis coding index | 7.30 | — |
| artificial analysis intelligence index | 12.40 | 17.20 |
| artificial analysis math index | — | 63.00 |
| gpqa | 43.3% | 61.5% |
| hle | 3.8% | 5.5% |
| ifbench | 38.6% | 22.9% |
| lcr | 12.0% | 9.7% |
| livecodebench | — | 27.0% |
| math 500 | — | 94.1% |
| mmlu pro | — | 73.9% |
| scicode | 21.8% | 37.6% |
| tau2 | 27.8% | — |
| terminalbench hard | 0.0% | — |
Benchmark data from Artificial Analysis.