MEGA-Bench Task - Counterfactual Arithmetic: leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 3 27B (IT) | 100 |
| 2 | Gemini 2.5 Pro (Preview 03-25) | 100 |
| 3 | GPT-4o Mini | 92.9 |
| 4 | GPT-4o | 85.7 |
| 5 | InternVL3-8B | 85.7 |
| 6 | Gemini 1.5 Pro (002) | 85.7 |
| 7 | Llama 4 Scout Base | 85.7 |
| 8 | Claude 3.5 Sonnet (20240620) | 78.6 |
| 9 | Gemma 3 4B (IT) | 42.9 |
| 10 | Qwen 2 VL 72B | 35.7 |
| 11 | Gemini 1.5 Flash (002) | 35.7 |
| 12 | Gemma 3 12B (IT) | 21.4 |
| 13 | Claude 3.5 Sonnet (20241022) | 14.3 |
| 14 | Aquila-VL-2B | 14.3 |
| 15 | InternVL2.5-78B | 14.3 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-counterfactual-arithmetic · How It Works · Data refreshed daily, snapshot 2026-09-05.