AraGen: leaderboard
Arabic generative benchmark using the six-dimension 3C3H metric across QA, grammar, reasoning, and safety tasks. This board uses the v3 question set and Claude Sonnet 4 judge, preserving the published point estimates and distinct precision configurations.
Metric: 3C3H Score (%). Source: huggingface.co. Saturation forecast: Around January 2027. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 (2024-12-17) | 84.29 | #149 |
| 2 | GPT-5 | 84.25 | #98 |
| 3 | O3 (2025-04-16) | 82.19 | #121 |
| 4 | Claude Opus 4 (20250514) | 80.96 | #154 |
| 5 | Claude Opus 4.5 (20251101) | 80.29 | #74 |
| 6 | Claude Sonnet 4.5 | 78.17 | #142 |
| 7 | Claude 3.7 Sonnet (20250219) | 78.16 | #201 |
| 8 | Claude Sonnet 4 (20250514) | 75.58 | #213 |
| 9 | GPT-4.1 | 74.54 | #248 |
| 10 | Gemma 4 31B (IT) | 72.71 | #176 |
| 11 | O4 Mini (2025-04-16) | 70.6 | #179 |
| 12 | O1 Mini (2024-09-12) | 70.46 | #403 |
| 13 | Gemini 2.5 Pro (Preview 05-06) | 68.37 | #147 |
| 14 | Gemma 4 26B A4B (IT) | 67.35 | #297 |
| 15 | Claude Haiku 4.5 (20251001) | 65.8 | #261 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aragen · How It Works · Data refreshed daily, snapshot 2026-10-20.