Claude Opus 5 scores 63.9 on the Respan Index (#7), ahead of Gemini 3.8 Flash at 52.2 (#16). Claude Opus 5 leads on 18 of the 27 benchmarks both report. Gemini 3.8 Flash is 6.7x cheaper per token at list price. Gemini 3.8 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) |
| 85.6% |
| 68.4% |
| FrontierSWE V2 (Proximal Labs) | 52% | 19.6% |
| Furniture Assembly (Epoch AI) | 60.8% | 31.7% |
| GPQA Diamond | 93.9% | 95.3% |
| LiveBench Agentic Coding (LiveBench) | 65.2% | 54.2% |
| LiveBench Coding (LiveBench) | 81.4% | 72.5% |
| LiveBench Data Analysis (LiveBench) | 74.6% | 54% |
| LiveBench Instruction Following (LiveBench) | 63.8% | 81.4% |
| LiveBench Language (LiveBench) | 88.7% | 87.8% |
| LiveBench Mathematics (LiveBench) | 95.7% | 91.6% |
| LiveBench Reasoning (LiveBench) | 91.2% | 89.3% |
| Mystery Game Puzzles (Epoch AI) | 59% | 47% |
| OTIS Mock AIME 2024-2025 (Epoch AI) | 98.9% | 98.9% |
| SAGE (Vals AI) | 49.4% | 35.1% |
| SimpleQA Verified | 59.9% | 69.7% |
| Terminal-Bench 4.0 (Vals AI) | 53.5% | 19.2% |
| Vending-Bench 2 (Andon Labs) | 11181.87 | 5093.79 |
| WeirdML (Håvard Tveit Ihle) | 91.8% | 84.8% |
| LMArena Agent (LMArena) | 0.0867 | 0.0297 |
| LMArena Elo | 1490 | 1495 |
| LMArena Vision (LMArena) | 1289 | 1290 |
| LMArena WebDev (LMArena) | 1695 | 1583 |
| SimpleBench (SimpleBench) | 80.6% | 82.4% |
Higher is better. Each score links to where it was published.