LLM Stats (Multi-Challenge): leaderboard
Multi-step challenge problems requiring integration of reasoning across multiple domains and sustained logical chains.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 32 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Nova 2.0 Lite | 76.6 | #417 |
| 2 | GPT-5 | 69.6 | #98 |
| 3 | Qwen 3.5 397B A17B | 67.6 | #146 |
| 4 | Seed 1.8 | 66.7 | #139 |
| 5 | Nemotron 3 Ultra | 63.8 | #235 |
| 6 | Qwen 3 Max (Thinking) | 63.3 | #208 (Qwen 3 Max) |
| 7 | Step3 VL 10B | 62.6 | #485 |
| 8 | Qwen 3.5 122B A10B | 61.5 | #173 |
| 9 | Seed 2.0 Mini | 61.1 | #276 |
| 10 | Qwen 3.5 27B | 60.8 | #187 |
| 11 | O3 | 60.4 | #124 |
| 12 | Qwen 3.5 35B A3B | 60 | #256 |
| 13 | Nemotron 3 Super | 55.23 | #335 |
| 14 | Qwen 3.5 9B | 54.5 | #379 |
| 15 | Kimi K2 | 54.1 | #243 |
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-challenge · How It Works · Data refreshed daily, snapshot 2026-10-20.