LA Leaderboard - MGSM Catalan: leaderboard
MGSM grade-school maths word problems in Catalan, answered directly without chain of thought. From the LA Leaderboard, which evaluates models on Spanish, Catalan, Basque and Galician tasks with the lm-evaluation-harness; exact match after whitespace removal (%), higher is better. Each model result file is a separate run.
Metric: Exact Match (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 7B | 25.6 | #895 |
| 2 | Gemma 2 9B | 20 | #924 |
| 3 | Qwen3-8B-Base | 14 | #626 |
| 4 | SmolLM3-3B-Base | 13.6 | #1260 |
| 5 | Qwen 2.5 3B | 12.8 | #1177 |
| 6 | Yi-1.5-9B | 12.4 | #1215 |
| 7 | Qwen 2.5 1.5B | 9.6 | #1412 |
| 8 | Llama 3.1 8B | 9.2 | #1130 |
| 9 | gemma-3-4B-pt Base | 8.8 | #1110 |
| 10 | EuroLLM-9B | 8.4 | #1238 |
| 11 | Yi-1.5-9B Chat | 7.6 | #1032 |
| 12 | OLMo-2-1124-7B | 7.2 | #1381 |
| 13 | Mistral-7B-v0.3 | 6.4 | #1321 |
| 14 | salamandra-7B | 5.6 | #1446 |
| 15 | Llama 3.2 3B | 5.2 | #1376 |
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-mgsm-catalan · How It Works · Data refreshed daily, snapshot 2026-11-05.