RoboRMBench - Lexical Substitution: leaderboard
Metric: Score Crossing Rate (0-1). Source: arxiv.org. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 0.59 |
| 2 | Llama 4 Scout | 0.48 |
| 3 | Claude Haiku 4.5 | 0.48 |
| 4 | GPT-4o | 0.34 |
| 5 | Llama 4 Maverick | 0.27 |
| 6 | Gemini 3 Flash | 0.27 |
| 7 | Gemma 3 27B | 0.18 |
| 8 | GPT-5.1 | 0.15 |
| 9 | Qwen 3 VL 8B | 0.15 |
| 10 | Gemma 3 12B | 0.12 |
| 11 | Claude Sonnet 4.6 | 0.05 |
Interactive version: theaggregate.ai/benchmark?slug=robormbench-lexical-substitution · How It Works · Data refreshed daily, snapshot 2026-09-21.