Qwen3.8-Max scores 50.3 on the Respan Index (#22), ahead of Grok 4.5 at 43.2 (#38). Qwen3.8-Max leads on 19 of the 23 benchmarks both report. Both cost about the same per token at list price. Qwen3.8-Max has the larger context window (1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| 57.2% |
| 74.7% |
| Furniture Assembly (Epoch AI) | 22.5% | 20% |
| GPQA Diamond | 93.4% | 92.6% |
| LiveBench Agentic Coding (LiveBench) | 56.5% | 64.6% |
| LiveBench Coding (LiveBench) | 68.6% | 72.9% |
| LiveBench Data Analysis (LiveBench) | 73% | 78.4% |
| LiveBench Instruction Following (LiveBench) | 71.5% | 74.1% |
| LiveBench Language (LiveBench) | 82.8% | 79.7% |
| LiveBench Mathematics (LiveBench) | 90.8% | 91.3% |
| LiveBench Reasoning (LiveBench) | 87.2% | 88.2% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 97.8% | 99.4% |
| SAGE (Vals AI) | 35% | 51.3% |
| SimpleQA Verified | 48.3% | 45.8% |
| Terminal-Bench 4.0 (Vals AI) | 8.6% | 34.3% |
| LMArena Agent (LMArena) | 0.0121 | 0.0251 |
| LMArena Elo | 1450 | 1481 |
| LMArena Vision (LMArena) | 1279 | 1301 |
| LMArena WebDev (LMArena) | 1552 | 1671 |
| tau2-bench Banking Knowledge (Sierra) | 47.9% | 55.2% |
Higher is better. Each score links to where it was published.