MEGA-Bench Task - Defeasible Reasoning: leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 88.3 |
| 2 | Claude 3.5 Sonnet (20240620) | 85.5 |
| 3 | Gemini 1.5 Flash (002) | 84.5 |
| 4 | Gemma 3 12B (IT) | 83.8 |
| 5 | Claude 3.5 Sonnet (20241022) | 83.8 |
| 6 | GPT-4o | 82.8 |
| 7 | InternVL3-78B | 82.8 |
| 8 | Llama 4 Scout Base | 82.1 |
| 9 | GPT-4o Mini | 81.7 |
| 10 | Gemma 3 27B (IT) | 81 |
| 11 | Gemini 1.5 Pro (002) | 79.7 |
| 12 | InternVL3-38B | 77.6 |
| 13 | Gemma 3 4B (IT) | 72.4 |
| 14 | InternVL2.5-78B | 72.4 |
| 15 | Qwen 2 VL 72B | 71.7 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-defeasible-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.