NeoEvalPlusN: leaderboard
Public leaderboard for proprietary command-following, distractor-resistance, expectation-breaking, poem, and stylized-writing tests run mainly on open-source LLM variants.
Metric: Mean Score (0-6). Source: huggingface.co. Saturation forecast: Around January 2027. 203 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V3 | 3.7 | #321 |
| 2 | Kimi K2 0905 | 3.2 | #288 |
| 3 | Kimi K2 | 3.1 | #243 |
| 4 | DeepSeek V2 Chat (0628) | 2.8 | #400 |
| 5 | GLM-4 32B (0414) | 2.35 | #533 |
| 6 | Llama 4 Scout Instruct | 2.05 | #539 |
| 7 | Gemma 3 27B (IT) | 2 | #523 |
| 8 | Gemma 2 27B (IT) | 2 | #645 |
| 9 | Gemma 2 9B (IT) | 1.95 | #788 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=neoevalplusn · How It Works · Data refreshed daily, snapshot 2026-10-20.