Hebrew LLM - Nikud — leaderboard
Metric: Score (%). Source: huggingface.co. 30 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 88.37 |
| 2 | Gemini 3 Pro (Preview) | 87.58 |
| 3 | Gemini 2.5 Pro | 87.15 |
| 4 | GPT-5.1 | 80.55 |
| 5 | Gemini 2.5 Flash | 79.52 |
| 6 | Qwen 3.5 397B A17B | 78.23 |
| 7 | DeepSeek V3.1 | 75.29 |
| 8 | GPT-4o | 74.82 |
| 9 | Gemma 4 31B (IT) | 73.41 |
| 10 | Gemini 2.5 Flash Lite | 70.73 |
| 11 | Gemma 4 26B A4B (IT) | 68.07 |
| 12 | Qwen 3.5 122B A10B | 67.11 |
| 13 | GPT-5 Mini | 66.22 |
| 14 | Claude Sonnet 4.5 | 62.95 |
| 15 | Claude Haiku 4.5 | 62.53 |
Interactive version: aibenchmarks.dev/benchmark?slug=hebrew-llm-nikud · How the rankings work · Data refreshed daily, snapshot 2026-07-12.