EVALITA - lexical-substitution — leaderboard
Metric: CPS. Source: huggingface.co. 48 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 Next 80B A3B Instruct | 54.53 |
| 2 | Mistral Small 3 | 52.67 |
| 3 | Qwen 2.5 72B Instruct | 50.24 |
| 4 | Llama 3.3 70B Instruct | 46.8 |
| 5 | Gemma 3 27B (IT) | 42.41 |
| 6 | Lexora-Medium-7B | 40.88 |
| 7 | Qwen 2.5 VL 32B Instruct | 40.67 |
| 8 | Llama 4 Scout Instruct | 39.29 |
| 9 | calme-3.2-instruct-78B | 38.82 |
| 10 | DeepSeek R1 Distill Llama 70B | 38.61 |
| 11 | Mistral Large 2 (Nov) Instruct (2411) | 38.56 |
| 12 | Qwen2.5-14B-Instruct-1M | 36.76 |
| 13 | Gemma 2 27B (IT) | 36.04 |
| 14 | Phi-3-medium-4k-instruct | 35.87 |
| 15 | Phi-4 | 35.1 |
Interactive version: theaggregate.ai/benchmark?slug=evalita-lexical-substitution · How the rankings work · Data refreshed daily, snapshot 2026-07-22.