DeepSeek V3.1-Terminus scores 18.7 on the Respan Index (#99), ahead of Qwen3 235B-A22B Instruct 2507 at 18.5 (#101). DeepSeek V3.1-Terminus leads on 2 of the 3 benchmarks both report. Qwen3 235B-A22B Instruct 2507 has the larger context window (262K tokens).
| Benchmark | DeepSeek V3.1-Terminus | Qwen3 235B-A22B Instruct 2507 |
|---|---|---|
| LMArena Elo | 1417 | 1422 |
| Aider-Polyglot | 76.1% | 57.3% |
| MMLU-Pro | 85% | 83% |
Higher is better. Each score links to where it was published.
* Estimated: no published scores in that category. How the Respan Index works