LA Leaderboard: leaderboard
Official LA Leaderboard average over 65 chance-normalized tasks: first average within Spanish, Catalan, Basque and Galician, then weight the four languages equally. Preserves the source treatment of missing task metrics.
Metric: Normalized Four-language Average (%). Source: huggingface.co. Saturation forecast: Around August 2028. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 2 9B (IT) | 33.62 | #788 |
| 2 | Gemma 2 9B | 32.97 | #958 |
| 3 | Llama 3.1 8B Instruct | 30.23 | #1035 |
| 4 | Qwen 2.5 7B | 27.61 | #922 |
| 5 | Llama 3.1 8B | 27.04 | #1151 |
| 6 | EuroLLM-9B | 25.87 | #1254 |
| 7 | gemma-3-4B-pt Base | 25.85 | #1134 |
| 8 | Gemma 3 4B (IT) | 25.38 | #992 |
| 9 | salamandra-7B-instruct | 25.13 | #1411 |
| 10 | EuroLLM-9B-Instruct | 24.46 | #1054 |
| 11 | aya-expanse-8B | 24.3 | #931 |
| 12 | salamandra-7B | 24.04 | #1461 |
| 13 | Mistral 7B Instruct (v0.3) | 23.86 | #1274 |
| 14 | Qwen 2.5 7B Instruct | 23.66 | #862 |
| 15 | Yi-1.5-9B | 23.35 | #1241 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-20.