Hebrew LLM - Translation (Pairwise) — leaderboard

Metric: Pairwise Score (%). Source: huggingface.co. 30 models tracked.

Top models

#ModelScore
1Gemini 3 Pro (Preview)51.15
2Gemini 3 Flash (Preview)50.65
3Gemini 2.5 Pro50.45
4GPT-5.144.23
5Qwen 3.5 397B A17B42.88
6GPT-5 Mini41.52
7GLM-5.137.41
8Gemini 2.5 Flash35.01
9Claude Sonnet 4.533.8
10Gemma 4 31B (IT)33.3
11GPT-4o32.05
12Qwen 3.5 122B A10B30.99
13GLM-4.630.04
14Gemma 4 26B A4B (IT)27.28
15Gemma 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.