Grok 4.6 scores 51.0 on the Respan Index (#19), ahead of Qwen3.8-Max at 50.3 (#22). Qwen3.8-Max leads on 13 of the 24 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
| 66% |
| 74.7% |
| FrontierSWE V2 (Proximal Labs) | 25.3% | 17.8% |
| Furniture Assembly (Epoch AI) | 40% | 20% |
| GPQA Diamond | 94% | 92.6% |
| LiveBench Agentic Coding (LiveBench) | 57% | 64.6% |
| LiveBench Coding (LiveBench) | 76.8% | 72.9% |
| LiveBench Data Analysis (LiveBench) | 73.9% | 78.4% |
| LiveBench Instruction Following (LiveBench) | 71.9% | 74.1% |
| LiveBench Language (LiveBench) | 83.7% | 79.7% |
| LiveBench Mathematics (LiveBench) | 92.6% | 91.3% |
| LiveBench Reasoning (LiveBench) | 90.5% | 88.2% |
| Mystery Game Puzzles (Epoch AI) | 34% | 38% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 99.2% | 99.4% |
| SAGE (Vals AI) | 28.9% | 51.3% |
| SimpleQA Verified | 49.3% | 45.8% |
| Terminal-Bench 4.0 (Vals AI) | 17.2% | 34.3% |
| LMArena Agent (LMArena) | 0.0128 | 0.0251 |
| LMArena Elo | 1454 | 1481 |
| LMArena Vision (LMArena) | 1264 | 1301 |
| LMArena WebDev (LMArena) | 1620 | 1671 |
Higher is better. Each score links to where it was published.