NovGauge - Pairwise Novelty - Task Verified F1: leaderboard
Metric: Verified F1 (%). Source: arxiv.org. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 71.5 |
| 2 | GPT-5.4 Mini | 52.1 |
| 3 | Gemma 4 31B (IT) | 34.9 |
| 4 | Gemini 3.1 Pro (Preview) | 33.5 |
| 5 | GLM-5.1 | 24 |
| 6 | Gemini 3.1 Flash Lite | 22.8 |
| 7 | Qwen 3.6 Plus | 22 |
| 8 | DeepSeek V4 Pro | 20 |
| 9 | Kimi K2.6 | 19.9 |
| 10 | Claude Sonnet 4.6 | 18.4 |
| 11 | Claude Opus 4.7 | 18.2 |
| 12 | DeepSeek V4 Flash | 18.1 |
| 13 | Qwen 3 32B | 14.5 |
| 14 | Qwen 3.5 27B | 14.3 |
| 15 | MiniMax-M2.7 | 9.9 |
Interactive version: theaggregate.ai/benchmark?slug=novgauge-pairwise-novelty-task-verified-f1 · How It Works · Data refreshed daily, snapshot 2026-09-19.