LA Leaderboard - AQuAS: leaderboard
Raw semantic answer similarity for AQuAS, scaled from -1..1 to -100..100; higher is better. This task board is not the chance-normalized contribution to the LA aggregate.
Metric: Semantic Answer Similarity (x100). Source: huggingface.co. Saturation forecast: Estimated already saturated. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral 7B Instruct (v0.3) | 73.27 | #1274 |
| 2 | Gemma 2 9B (IT) | 71.86 | #788 |
| 3 | occiglot-7B-eu5-instruct | 71.43 | |
| 4 | Qwen 2.5 7B | 70.73 | #922 |
| 5 | Llama 3.1 8B Instruct | 70.62 | #1035 |
| 6 | Qwen 2.5 7B Instruct | 69.85 | #862 |
| 7 | Llama 3.2 3B Instruct | 69.77 | #1352 |
| 8 | EuroLLM-9B-Instruct | 69.61 | #1054 |
| 9 | Gemma 2 2B (IT) | 68.53 | #1341 |
| 10 | salamandra-7B-instruct | 68.26 | #1411 |
| 11 | Gemma 3 4B (IT) | 68.24 | #992 |
| 12 | Qwen 2.5 3B | 67.79 | #1211 |
| 13 | Qwen 2.5 3B Instruct | 67.69 | #1162 |
| 14 | Yi-1.5-9B Chat | 67.34 | #1031 |
| 15 | aya-expanse-8B | 66.91 | #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-aquas · How It Works · Data refreshed daily, snapshot 2026-10-20.