LLM Stats (Multi-IF): leaderboard
4,501 multi-turn conversations in 8 languages testing the ability to follow increasingly complex nested instructions across successive dialogue turns.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Ling-3.0-flash | 87.7 | #206 |
| 2 | Qwen 3 235B A22B 2507 (Thinking) | 80.6 | #260 (Qwen 3 235B A22B 2507) |
| 3 | LFM2.5-2.6B | 80.07 | #822 |
| 4 | O3 Mini | 79.5 | #272 |
| 5 | Qwen 3 VL 235B A22B (Thinking) | 79.1 | #238 (Qwen 3 VL 235B A22B) |
| 6 | Qwen 3 VL 32B (Thinking) | 78 | #294 (Qwen 3 VL 32B) |
| 7 | Qwen 3 Next 80B A3B (Thinking) | 77.8 | #320 (Qwen 3 Next 80B A3B) |
| 8 | Qwen 3 235B A22B 2507 Instruct | 77.5 | #300 |
| 9 | Qwen 3 VL 235B A22B Instruct | 76.3 | #273 |
| 10 | Qwen 3 Next 80B A3B Instruct | 75.8 | #368 |
| 11 | Qwen 3 VL 8B Instruct | 75.1 | #420 |
| 12 | Qwen 3 VL 8B (Thinking) | 75.1 | |
| 13 | Qwen 3 14B | 74.8 | #537 |
| 14 | Qwen 3 VL 4B (Thinking) | 73.6 | #480 (Qwen 3 VL 4B) |
| 15 | Qwen 3 VL 30B A3B (Thinking) | 73 | #351 (Qwen 3 VL 30B A3B) |
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-multi-if · How It Works · Data refreshed daily, snapshot 2026-10-20.