LA Leaderboard - GalCoLA: leaderboard
Raw Matthews correlation coefficient for GalCoLA linguistic acceptability, scaled to -100..100; higher is better. This is MCC, not classification accuracy.
Metric: MCC (x100). Source: huggingface.co. Saturation forecast: Around 2030. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 2 9B (IT) | 33.3 | #788 |
| 2 | Qwen 2.5 7B Instruct | 22.54 | #862 |
| 3 | Qwen 3 8B | 20.1 | #674 |
| 4 | EuroLLM-9B-Instruct | 20.07 | #1054 |
| 5 | EuroLLM-9B | 19.72 | #1254 |
| 6 | Qwen 2.5 3B Instruct | 16.12 | #1162 |
| 7 | Gemma 2 9B | 15.25 | #958 |
| 8 | Llama 3.1 8B Instruct | 15.09 | #1035 |
| 9 | Qwen3-8B-Base | 13.64 | #626 |
| 10 | Gemma 3 4B (IT) | 12.07 | #992 |
| 11 | Qwen 2.5 7B | 12.06 | #922 |
| 12 | SmolLM3-3B | 10.18 | #1252 |
| 13 | Yi-1.5-9B | 8.08 | #1241 |
| 14 | Llama 3.1 8B | 8.01 | #1151 |
| 15 | aya-expanse-8B | 7.91 | #931 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=la-leaderboard-galcola · How It Works · Data refreshed daily, snapshot 2026-10-20.