DeepSeek V4 Flash scores 43.3 on the Respan Index (#37), ahead of Qwen3.7-Max at 40.5 (#45). DeepSeek V4 Flash leads on 9 of the 16 benchmarks both report.
* Estimated: no published scores in that category. How the Respan Index works
| LiveBench Agentic Coding (LiveBench) |
| 46.8% |
| 43.6% |
| LiveBench Coding (LiveBench) | 75% | 74.2% |
| LiveBench Data Analysis (LiveBench) | 79.3% | 71.8% |
| LiveBench Instruction Following (LiveBench) | 65.5% | 74% |
| LiveBench Language (LiveBench) | 79.2% | 79.7% |
| LiveBench Mathematics (LiveBench) | 86.8% | 85.2% |
| LiveBench Reasoning (LiveBench) | 86.6% | 83.3% |
| Mystery Game Puzzles (Epoch AI) | 34% | 32% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 94.4% | 95.6% |
| SimpleQA Verified | 33.6% | 55.8% |
| LMArena WebDev (LMArena) | 1581 | 1515 |
| SimpleBench (SimpleBench) | 61.1% | 70.4% |
Higher is better. Each score links to where it was published.