← All comparisons
DeepSeek V4 Pro (Reasoning, Max Effort) vs Nova 2.0 Pro Preview (low)
DeepSeek vs Amazon — side-by-side benchmark comparison
| DeepSeek V4 Pro (Reasoning, Max Effort) | Nova 2.0 Pro Preview (low) | |
|---|---|---|
| Intelligence Index | 51.5 | 31.9 |
| Coding Index | 47.5 | 24.5 |
| Math Index | — | 63.3 |
| Output speed (tok/s) | 51.8 | 129.1 |
| Blended price ($/1M) | $0.54 | $3.44 |
| Time to first token (s) | 1.21s | 3.95s |
| aime | — | — |
| aime 25 | — | 63.3% |
| artificial analysis coding index | 47.50 | 24.50 |
| artificial analysis intelligence index | 51.50 | 31.90 |
| artificial analysis math index | — | 63.30 |
| gpqa | 88.8% | 75.1% |
| hle | 35.9% | 5.2% |
| ifbench | 76.5% | 79.6% |
| lcr | 66.3% | 61.7% |
| livecodebench | — | 63.8% |
| math 500 | — | — |
| mmlu pro | — | 82.2% |
| scicode | 50.0% | 38.7% |
| tau2 | 96.2% | 90.6% |
| terminalbench hard | 46.2% | 17.4% |
Benchmark data from Artificial Analysis.