DeepSeek V3 scores 12.4 on the Respan Index (#132), ahead of Qwen3 30B-A3B Instruct 2507 at 11.5 (#134). Qwen3 30B-A3B Instruct 2507 leads on 3 of the 4 benchmarks both report. Qwen3 30B-A3B Instruct 2507 has the larger context window (262K tokens).
| Benchmark | DeepSeek V3 | Qwen3 30B-A3B Instruct 2507 |
|---|---|---|
| GPQA Diamond | 59.1% | 55.6% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 15.8% | 62.2% |
| LMArena Elo | 1332 | 1383 |
| MMLU-Pro | 75.9% | 78.4% |
Higher is better. Each score links to where it was published.
* Estimated: no published scores in that category. How the Respan Index works