Hebrew LLM - Translation (Pairwise): leaderboard
The Translation task of the Hebrew LLM Leaderboard — Hebrew translation quality judged in pairwise comparisons; pairwise score (%).
Metric: Pairwise Score (%). Source: huggingface.co. Saturation forecast: Around January 2027. 34 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 51.15 | #66 |
| 2 | Gemini 3 Flash (Preview) | 50.65 | #82 |
| 3 | Gemini 2.5 Pro | 50.45 | #152 |
| 4 | GPT-5.1 | 44.23 | #135 |
| 5 | Qwen 3.5 397B A17B | 42.88 | #146 |
| 6 | GPT-5 Mini | 41.52 | #186 |
| 7 | GLM-5.3 Flash | 40.27 | #65 |
| 8 | GLM-5.1 | 37.41 | #97 |
| 9 | Gemini 2.5 Flash | 35.01 | #245 |
| 10 | Claude Sonnet 4.5 | 33.8 | #142 |
| 11 | Gemma 4 31B (IT) | 33.3 | #176 |
| 12 | GPT-4o | 32.05 | #343 |
| 13 | Qwen 3.5 122B A10B | 30.99 | #173 |
| 14 | GLM-4.6 | 30.04 | #252 |
| 15 | MiniMax-M2.7 | 28.13 | #258 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=hebrew-llm-translation-pairwise · How It Works · Data refreshed daily, snapshot 2026-10-20.