ArabCulture-Dialogue - MSA to Dialect Translation: leaderboard
ArabCulture-Dialogue (MBZUAI, 2026) asks models to translate culturally grounded dialogues from Modern Standard Arabic into the dialect of a named Arab country, zero-shot with the target country and region named in the prompt. A GPT-5 judge rates the overall quality of each translation from 1 to 5; higher is better.
Metric: Overall translation quality (1-5), GPT-5 judge (low reasoning) holistic rating, zero-shot MSA to Dialect translation of the dialogues of ArabCulture-Dialogue (13 Arab countries, parallel Modern Standard Arabic and country-dialect versions of 3,471 culturally grounded multi-turn dialogues built from ArabCulture), with the target country and region named in the prompt; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 4.53 | #178 |
| 2 | Gemini 2.5 Pro | 4.19 | #276 |
| 3 | ALLaM-7B-Instruct-preview | 3.41 | #2001 |
| 4 | c4ai-command-r7B-arabic-02-2025 | 3 | #2135 |
| 5 | Fanar-1-9B | 2.94 | #1913 |
| 6 | jais-adapted-7B-chat | 2.44 | #5831 |
| 7 | Qwen 3 8B | 2.35 | #1873 |
| 8 | Hala-9B | 1.76 | #1162 |
| 9 | Gemma 2 9B (IT) | 1.5 | #2369 |
| 10 | Llama 3.1 8B Instruct | 1.34 | #4600 |
| 11 | SILMA-9B-Instruct-v1.0 | 1.3 | #3847 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=arabculture-dialogue-msa-to-dialect-translation · How It Works · Data refreshed daily, snapshot 2026-11-21.