CoS-VQ BF16 - HellaSwag: leaderboard
The CoS-VQ study reports these reference results for unquantized models on language modeling and zero-shot questions. Lower perplexity is better. Zero-shot accuracy (%).
Metric: Zero-shot accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.1 70B Instruct | 84.63 | #559 |
| 2 | Qwen 3 32B | 82.58 | #443 |
| 3 | Llama 3.1 8B Instruct | 79.17 | #1035 |
| 4 | Qwen 3 14B | 78.86 | #536 |
| 5 | Qwen 3 8B | 74.94 | #671 |
| 6 | Qwen 3 4B | 68.43 | #833 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cos-vq-bf16-hellaswag · How It Works · Data refreshed daily, snapshot 2026-10-29.