GPT-6 Sol scores 62.8 on the Respan Index (#8), ahead of Gemini 3.8 Flash at 52.2 (#16). GPT-6 Sol leads on 13 of the 23 benchmarks both report. Gemini 3.8 Flash is 2.7x cheaper per token at list price. GPT-6 Sol has the larger context window (1.1M tokens).
* Estimated: no published scores in that category. How the Respan Index works
| GPQA Diamond |
| 95.3% |
| 94.3% |
| LiveBench Agentic Coding (LiveBench) | 54.2% | 52.9% |
| LiveBench Coding (LiveBench) | 72.5% | 81.8% |
| LiveBench Data Analysis (LiveBench) | 54% | 81.2% |
| LiveBench Instruction Following (LiveBench) | 81.4% | 68.6% |
| LiveBench Language (LiveBench) | 87.8% | 85.3% |
| LiveBench Mathematics (LiveBench) | 91.6% | 96.4% |
| LiveBench Reasoning (LiveBench) | 89.3% | 88.7% |
| Mystery Game Puzzles (Epoch AI) | 47% | 56% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 98.9% | 100% |
| SAGE (Vals AI) | 35.1% | 44.8% |
| SimpleQA Verified | 69.7% | 60.7% |
| Terminal-Bench 4.0 (Vals AI) | 19.2% | 44.4% |
| Vending-Bench 2 (Andon Labs) | 5093.79 | 14427.85 |
| LMArena Agent (LMArena) | 0.0297 | 0.0971 |
| LMArena Elo | 1495 | 1457 |
| LMArena WebDev (LMArena) | 1583 | 1689 |
| SimpleBench (SimpleBench) | 82.4% | 73.1% |
| DeepSWE v1.1 | 73.7% | 68.8% |
Higher is better. Each score links to where it was published.