Vibe-Bench - BM25 Retrieval - Preference Accuracy: 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 | 59.38 |
| 2 | Llama 3.1 70B Instruct | 57.03 |
| 3 | Mistral Small 3.2 | 57.03 |
| 4 | Llama 3.1 8B Instruct | 53.91 |
| 5 | Gemma 3 27B (IT) | 53.91 |
| 6 | Qwen 3 14B | 51.56 |
Interactive version: theaggregate.ai/benchmark?slug=vibe-bench-bm25-retrieval-preference-accuracy · How It Works · Data refreshed daily, snapshot 2026-09-24.