← All comparisons
Nova 2.0 Lite (medium) vs DeepSeek V3.2 Exp (Reasoning)
Amazon vs DeepSeek — side-by-side benchmark comparison
| Nova 2.0 Lite (medium) | DeepSeek V3.2 Exp (Reasoning) | |
|---|---|---|
| Intelligence Index | 29.7 | 32.9 |
| Coding Index | 23.9 | 33.3 |
| Math Index | 88.7 | 87.7 |
| Output speed (tok/s) | 168.5 | 0.0 |
| Blended price ($/1M) | $0.85 | $0.31 |
| Time to first token (s) | 13.49s | 0.00s |
| aime | — | — |
| aime 25 | 88.7% | 87.7% |
| artificial analysis coding index | 23.90 | 33.30 |
| artificial analysis intelligence index | 29.70 | 32.90 |
| artificial analysis math index | 88.70 | 87.70 |
| gpqa | 76.8% | 79.7% |
| hle | 8.6% | 13.8% |
| ifbench | 68.5% | 54.1% |
| lcr | 58.3% | 69.0% |
| livecodebench | 66.3% | 78.9% |
| math 500 | — | — |
| mmlu pro | 81.3% | 85.0% |
| scicode | 36.8% | 37.7% |
| tau2 | 75.7% | 33.9% |
| terminalbench hard | 17.4% | 31.1% |
Benchmark data from Artificial Analysis.