NOLLI - Average: leaderboard
Metric: Accuracy (%; 25-task macro average). Source: arxiv.org. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (Medium) | 84.9 |
| 2 | Claude Opus 4.8 (Medium) | 78.3 |
| 3 | Gemini 3.1 Pro (Preview) (Medium) | 73.6 |
| 4 | Qwen 3.5 397B A17B (Non-reasoning) | 62.5 |
| 5 | Gemma 4 31B (IT) (Thinking) | 48.3 |
| 6 | GPT-OSS-120B (Medium) | 47.7 |
| 7 | DeepSeek V4 Flash | 45.8 |
| 8 | Qwen 3.5 27B | 44.4 |
| 9 | Qwen 3.5 9B (Non-reasoning) | 32.5 |
| 10 | EXAONE-4.0-32B (Thinking) | 25.1 |
| 11 | Llama 4 Maverick | 14.4 |
| 12 | Solar Open 100B | 14.1 |
| 13 | Llama 3.1 8B Instruct | 1.5 |
Interactive version: theaggregate.ai/benchmark?slug=nolli-average · How It Works · Data refreshed daily, snapshot 2026-09-19.