Qwen3.8-Max scores 50.3 on the Respan Index (#22), ahead of Gemini 3.7 Flash at 48.6 (#26). They split the 24 benchmarks both report evenly. Gemini 3.7 Flash is 2.0x cheaper per token at list price. Gemini 3.7 Flash has the larger context window (1.0M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| FrontierMath Tiers 1-3 v2 (Epoch AI) |
| 71.6% |
| 74.7% |
| FrontierSWE V2 (Proximal Labs) | 20.3% | 17.8% |
| Furniture Assembly (Epoch AI) | 26.7% | 20% |
| GPQA Diamond | 94.8% | 92.6% |
| LiveBench Agentic Coding (LiveBench) | 58.3% | 64.6% |
| LiveBench Coding (LiveBench) | 78.9% | 72.9% |
| LiveBench Data Analysis (LiveBench) | 68% | 78.4% |
| LiveBench Instruction Following (LiveBench) | 79.9% | 74.1% |
| LiveBench Language (LiveBench) | 85.5% | 79.7% |
| LiveBench Mathematics (LiveBench) | 93.5% | 91.3% |
| LiveBench Reasoning (LiveBench) | 87.8% | 88.2% |
| Mystery Game Puzzles (Epoch AI) | 37% | 38% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 97.2% | 99.4% |
| SAGE (Vals AI) | 49.2% | 51.3% |
| SimpleQA Verified | 69.2% | 45.8% |
| Terminal-Bench 4.0 (Vals AI) | 12.1% | 34.3% |
| LMArena Agent (LMArena) | -0.0149 | 0.0251 |
| LMArena Elo | 1488 | 1481 |
| LMArena Vision (LMArena) | 1296 | 1301 |
| LMArena WebDev (LMArena) | 1592 | 1671 |
Higher is better. Each score links to where it was published.