Mizan LLM Leaderboard: leaderboard
Persian/Farsi evaluation across Persian IFEval, MT-Bench, PerMMLU, PerCoR, Persian NLU, and Persian NLG. Preserves the official Average scaled to 0-100 and explicit source thinking-mode settings; higher is better.
Metric: Average Score (0-100). Source: huggingface.co. Saturation forecast: Estimated already saturated. 45 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 | 74.68 | #124 |
| 2 | Gemini 2.5 Pro | 73.32 | #152 |
| 3 | GPT-5 Mini | 71.69 | #186 |
| 4 | GPT-4.1 | 69.92 | #248 |
| 5 | Gemini 2.5 Flash | 69.86 | #245 |
| 6 | Gemini 2.0 Flash | 68.86 | #342 |
| 7 | GPT-4o | 68.77 | #343 |
| 8 | Gemini 2.0 Flash Lite | 66.4 | #450 |
| 9 | GPT-5 Nano | 66.37 | #425 |
| 10 | GPT-OSS-120B | 66.33 | #341 |
| 11 | GPT-5 Mini (Minimal) | 65.6 | #186 (GPT-5 Mini) |
| 12 | GPT-4.1 Mini | 65.56 | #355 |
| 13 | DeepSeek Reasoner | 65.49 | #242 |
| 14 | DeepSeek V3 Chat | 64.58 | #447 |
| 15 | Gemma 3 27B (IT) | 62.47 | #523 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mizan-llm-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.