Vibe-Bench - BERT Retrieval - Preference Precision: leaderboard
Metric: Score (0-100). Source: arxiv.org. Saturation forecast: Around December 2027. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 235B A22B 2507 Instruct | 59.18 |
| 2 | Mistral Small 3.2 | 55.41 |
| 3 | Llama 3.1 70B Instruct | 54.88 |
| 4 | Qwen 3 14B | 54.79 |
| 5 | Llama 3.1 8B Instruct | 51.79 |
| 6 | Gemma 3 27B (IT) | 49.32 |
Interactive version: theaggregate.ai/benchmark?slug=vibe-bench-bert-retrieval-preference-precision · How It Works · Data refreshed daily, snapshot 2026-09-24.