DeepSeek V3.2 scores 24.1 on the Respan Index (#85), ahead of o3 at 23.9 (#86). DeepSeek V3.2 leads on 3 of the 4 benchmarks both report. o3 has the larger context window (200K tokens).
| Benchmark | DeepSeek V3.2 | o3 |
|---|---|---|
| LMArena Elo | 1425 | 1432 |
| AIME 2025 | 93.1% | 89.2% |
| GPQA Diamond | 82.4% | 81.8% |
| SWE-bench Verified | 73.1% | 69.1% |
Higher is better. Each score links to where it was published.
* Estimated: no published scores in that category. How the Respan Index works