LLM Stats (Wild Bench): leaderboard
1,024 challenging real-user tasks from chatbot logs testing instruction following on complex, unconstrained user requests.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Large 3 | 68.5 | #401 |
| 2 | Ministral-3-14B-Instruct-2512 | 68.5 | #615 |
| 3 | Ministral-3-8B-Instruct-2512 | 66.8 | #661 |
| 4 | Mistral Small 3.2 | 65.33 | #605 |
| 5 | Ministral-3-3B-Instruct-2512 | 56.8 | #855 |
| 6 | Mistral Small 3 | 52.2 | #627 |
| 7 | Jamba 1.5 Large | 48.5 | #669 |
| 8 | Jamba 1.5 Mini | 42.4 | #983 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-wild-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.