Hebrew LLM - Translation (Pairwise) — leaderboard
Metric: Pairwise Score (%). Source: huggingface.co. 30 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 51.15 |
| 2 | Gemini 3 Flash (Preview) | 50.65 |
| 3 | Gemini 2.5 Pro | 50.45 |
| 4 | GPT-5.1 | 44.23 |
| 5 | Qwen 3.5 397B A17B | 42.88 |
| 6 | GPT-5 Mini | 41.52 |
| 7 | GLM-5.1 | 37.41 |
| 8 | Gemini 2.5 Flash | 35.01 |
| 9 | Claude Sonnet 4.5 | 33.8 |
| 10 | Gemma 4 31B (IT) | 33.3 |
| 11 | GPT-4o | 32.05 |
| 12 | Qwen 3.5 122B A10B | 30.99 |
| 13 | GLM-4.6 | 30.04 |
| 14 | Gemma 4 26B A4B (IT) | 27.28 |
| 15 | Gemma 3 27B (IT) | 26.73 |
Interactive version: aibenchmarks.dev/benchmark?slug=hebrew-llm-translation-pairwise · How the rankings work · Data refreshed daily, snapshot 2026-07-12.