xbench-ScienceQA - BoN (N=5): leaderboard
The best-of-five (BoN, N=5) score xbench reports beside the five-run average on its graduate-level science questions (search off, gemini-3.5-flash judge), so it rewards a model that gets a question right in any of five tries; the same two ambiguous rows (GPT-6 Astra and Deepseek-V4-Flash) are held.
Metric: Best-of-5 score (%). Source: xbench.org. Saturation forecast: Not forecast. 28 models tracked. Updated 19 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Kimi K3 | 86 | #61 | |
| 2 | Gemini 3 Pro (Preview) | 85 | #120 | |
| 3 | Claude Opus 4.8 | 84 | #64 | |
| 4 | Claude Opus 5.5 | 83 | #9 | |
| 5 | Claude Fable 5 | 83 | #32 | |
| 6 | Qwen 3.8 Max | 82 | #66 | |
| 7 | DeepSeek V4.1 Flash | Max | 82 | #69 |
| 8 | Gemini 3.7 Flash | 81 | #42 | |
| 9 | DeepSeek V4 Pro | Max | 81 | #198 |
| 10 | Grok 4.5 | 81 | #90 | |
| 11 | Qwen 3.8 Max (Preview) | 81 | #80 | |
| 12 | GLM-5.3 | 80 | #73 | |
| 13 | Hy3 | 80 | #211 | |
| 14 | GPT-5.5 Pro | High | 80 | #40 |
| 15 | GPT-5.4 | 79 | #138 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=xbench-scienceqa-bon-n-5 · How It Works · Data refreshed daily, snapshot 2026-11-20.