BBEH: leaderboard
BIG-Bench Extra Hard: 27 tasks from Google DeepMind that push frontier models beyond standard BIG-Bench Hard.
Metric: Harmonic Mean (%). Source: github.com. Status: saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 Mini (High) | 44.8 |
| 2 | Gemini 2.0 Flash | 9.8 |
| 3 | Gemini 2.0 Flash Lite | 8 |
| 4 | DeepSeek R1 | 6.8 |
| 5 | GPT-4o | 6 |
| 6 | Gemma 3 27B | 4.9 |
| 7 | Gemma 3 12B | 4.5 |
| 8 | Gemma 2 27B (IT) | 4 |
| 9 | Llama 3.1 8B Instruct | 3.6 |
| 10 | Gemma 3 4B | 3.4 |
Interactive version: theaggregate.ai/benchmark?slug=bbeh · How It Works · Data refreshed daily, snapshot 2026-09-05.