Benchmark Almanac

MIZAN: A Persian LLM Leaderboard

o3 0.75 leads since May 2025
Not in model cards

Models answer questions, complete sentences, follow instructions, and hold conversations in Persian. The leaderboard averages six task scores.

The organizers publish overall scores for 45 model entries, alongside the six scores used to calculate each average.

Made by
MCINext
Released
May 2025
Kind
Knowledge in many languages
Level
School
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Persian
In and out
Text → Text
Interaction
Multi-turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
o3 #84 on the capability index
Last model added
September 2025
Scores refreshed
November 20th, 2026
41
Entries on its own leaderboard, 35 models we track

o3 leads with 0.75 among models released by September 2025 o3 leads with 0.75 among models released by September 2025

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in September 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 o3 Thinking 0.75
  2. 2 Gemini 2.5 Pro Jun 2025, Thinking 0.73
  3. 3 GPT-5 mini Thinking 0.72
  4. 4 Claude 3.7 Sonnet Non-thinking 0.71
  5. 5 GPT-4.1 Non-thinking 0.70
  6. 6 Gemini 2.5 Flash Sep 2025, Thinking 0.70
  7. 7 Gemini 2.0 Flash Feb 2025, Non-thinking 0.69
  8. 8 GPT-4o Mar 2025, Non-thinking 0.69
  9. 9 Gemini 2.0 Flash-Lite Non-thinking 0.66
  10. 10 GPT-5 nano Thinking 0.66
  11. 11 gpt-oss-120b Thinking 0.66
  12. 12 GPT-4.1 mini Non-thinking 0.66
  13. 13 deepseek-reasoner Thinking 0.65
  14. 14 deepseek-chat Non-thinking 0.65
  15. 15 Gemma 3 27B Non-thinking 0.62
  16. 16 GPT-4o mini Non-thinking 0.62
  17. 17 Qwen3-32B Non-thinking 0.62
  18. 18 gpt-oss-20b Thinking 0.62
  19. 19 Llama 3.3 70B Instruct Non-thinking 0.61
  20. 20 Cohere Command A Non-thinking 0.61
  21. 21 Gemma 3 12B Non-thinking 0.60
  22. 22 Qwen3-30B-A3B Non-thinking 0.59
  23. 23 Qwen3-14B Non-thinking 0.59
  24. 24 Llama 4 Scout Non-thinking 0.57
  25. 25 Command R+ Non-thinking 0.57
  26. 26 Mistral Small 3.1 Non-thinking 0.56
  27. 27 Claude 3.5 Haiku Non-thinking 0.55
  28. 28 GPT-4.1 nano Non-thinking 0.55
  29. 29 Gemma 3n E4B Non-thinking 0.54
  30. 30 Qwen3-8B Non-thinking 0.53
  31. 31 Gemma 3 4B Non-thinking 0.50
  32. 32 Aya Expanse 32B Non-thinking 0.49
  33. 33 deepseek-r1-0528-qwen3-8b Thinking 0.49
  34. 34 Command R (03-2024) Non-thinking 0.48
  35. 35 Qwen3-4B Non-thinking 0.48
  36. 36 yandexgpt-lite Non-thinking 0.42
  37. 37 Gemma 3 1B Non-thinking 0.33
  38. 38 Mistral-7B-Instruct-v0.3 Non-thinking 0.30
  39. 39 Llama-3.2-3B-Instruct Non-thinking 0.28
  40. 40 Llama-3.2-1B-Instruct Non-thinking 0.20
  41. 41 Gemma 3 270M Non-thinking 0.17

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

#EntryScoreSettingDateSource
1 o3 0.75 Thinking Apr 2025 Official leaderboard
2 Gemini 2.5 Pro Jun 2025 0.73 Thinking Jun 2025 Official leaderboard
3 GPT-5 mini 0.72 Thinking Aug 2025 Official leaderboard
4 Claude 3.7 Sonnet 0.71 Non-thinking Feb 2025 Official leaderboard
5 GPT-4.1 0.70 Non-thinking Apr 2025 Official leaderboard
6 Gemini 2.5 Flash Sep 2025 0.70 Thinking Sep 2025 Official leaderboard
7 Gemini 2.0 Flash Feb 2025 0.69 Non-thinking Feb 2025 Official leaderboard
8 GPT-4o Mar 2025 0.69 Non-thinking Mar 2025 Official leaderboard
9 Gemini 2.0 Flash-Lite 0.66 Non-thinking Feb 2025 Official leaderboard
10 GPT-5 nano 0.66 Thinking Aug 2025 Official leaderboard
11 gpt-oss-120b 0.66 Thinking Aug 2025 Official leaderboard
12 GPT-4.1 mini 0.66 Non-thinking Apr 2025 Official leaderboard
13 deepseek-reasoner open weights, 671B 0.65 Thinking Official leaderboard
14 deepseek-chat open weights, 671B 0.65 Non-thinking Official leaderboard
15 Gemma 3 27B 0.62 Non-thinking Mar 2025 Official leaderboard
16 GPT-4o mini 0.62 Non-thinking Jul 2024 Official leaderboard
17 Qwen3-32B 0.62 Non-thinking Apr 2025 Official leaderboard
18 gpt-oss-20b 0.62 Thinking Aug 2025 Official leaderboard
19 Llama 3.3 70B Instruct 0.61 Non-thinking Dec 2024 Official leaderboard
20 Cohere Command A 0.61 Non-thinking Mar 2025 Official leaderboard
21 Gemma 3 12B 0.60 Non-thinking Mar 2025 Official leaderboard
22 Qwen3-30B-A3B 0.59 Non-thinking Apr 2025 Official leaderboard
23 Qwen3-14B 0.59 Non-thinking Apr 2025 Official leaderboard
24 Llama 4 Scout 0.57 Non-thinking Apr 2025 Official leaderboard
25 Command R+ 0.57 Non-thinking Aug 2024 Official leaderboard
26 Mistral Small 3.1 0.56 Non-thinking Mar 2025 Official leaderboard
27 Claude 3.5 Haiku 0.55 Non-thinking Oct 2024 Official leaderboard
28 GPT-4.1 nano 0.55 Non-thinking Apr 2025 Official leaderboard
29 Gemma 3n E4B 0.54 Non-thinking Jun 2025 Official leaderboard
30 Qwen3-8B 0.53 Non-thinking Apr 2025 Official leaderboard
31 Gemma 3 4B 0.50 Non-thinking Mar 2025 Official leaderboard
32 Aya Expanse 32B 0.49 Non-thinking Oct 2024 Official leaderboard
33 deepseek-r1-0528-qwen3-8b 0.49 Thinking May 2025 Official leaderboard
34 Command R (03-2024) 0.48 Non-thinking Mar 2024 Official leaderboard
35 Qwen3-4B 0.48 Non-thinking Apr 2025 Official leaderboard
36 yandexgpt-lite open weights, 8B 0.42 Non-thinking Official leaderboard
37 Gemma 3 1B 0.33 Non-thinking Mar 2025 Official leaderboard
38 Mistral-7B-Instruct-v0.3 open weights, 7.3B 0.30 Non-thinking Official leaderboard
39 Llama-3.2-3B-Instruct open weights, 3.2B 0.28 Non-thinking Official leaderboard
40 Llama-3.2-1B-Instruct open weights, 1.2B 0.20 Non-thinking Official leaderboard
41 Gemma 3 270M 0.17 Non-thinking Aug 2025 Official leaderboard
#ModelScoreSettingDateSource
1 o3 0.75 Thinking Apr 2025 Official leaderboard
2 Gemini 2.5 Pro Jun 2025 0.73 Thinking Jun 2025 Official leaderboard
3 GPT-5 mini 0.72 Thinking Aug 2025 Official leaderboard
4 Claude 3.7 Sonnet 0.71 Non-thinking Feb 2025 Official leaderboard
5 GPT-4.1 0.70 Non-thinking Apr 2025 Official leaderboard
6 Gemini 2.5 Flash Sep 2025 0.70 Thinking Sep 2025 Official leaderboard
7 Gemini 2.0 Flash Feb 2025 0.69 Non-thinking Feb 2025 Official leaderboard
8 GPT-4o Mar 2025 0.69 Non-thinking Mar 2025 Official leaderboard
9 Gemini 2.0 Flash-Lite 0.66 Non-thinking Feb 2025 Official leaderboard
10 GPT-5 nano 0.66 Thinking Aug 2025 Official leaderboard
11 gpt-oss-120b 0.66 Thinking Aug 2025 Official leaderboard
12 GPT-4.1 mini 0.66 Non-thinking Apr 2025 Official leaderboard
13 Gemma 3 27B 0.62 Non-thinking Mar 2025 Official leaderboard
14 GPT-4o mini 0.62 Non-thinking Jul 2024 Official leaderboard
15 Qwen3-32B 0.62 Non-thinking Apr 2025 Official leaderboard
16 gpt-oss-20b 0.62 Thinking Aug 2025 Official leaderboard
17 Llama 3.3 70B Instruct 0.61 Non-thinking Dec 2024 Official leaderboard
18 Cohere Command A 0.61 Non-thinking Mar 2025 Official leaderboard
19 Gemma 3 12B 0.60 Non-thinking Mar 2025 Official leaderboard
20 Qwen3-30B-A3B 0.59 Non-thinking Apr 2025 Official leaderboard
21 Qwen3-14B 0.59 Non-thinking Apr 2025 Official leaderboard
22 Llama 4 Scout 0.57 Non-thinking Apr 2025 Official leaderboard
23 Command R+ 0.57 Non-thinking Aug 2024 Official leaderboard
24 Mistral Small 3.1 0.56 Non-thinking Mar 2025 Official leaderboard
25 Claude 3.5 Haiku 0.55 Non-thinking Oct 2024 Official leaderboard
26 GPT-4.1 nano 0.55 Non-thinking Apr 2025 Official leaderboard
27 Gemma 3n E4B 0.54 Non-thinking Jun 2025 Official leaderboard
28 Qwen3-8B 0.53 Non-thinking Apr 2025 Official leaderboard
29 Gemma 3 4B 0.50 Non-thinking Mar 2025 Official leaderboard
30 Aya Expanse 32B 0.49 Non-thinking Oct 2024 Official leaderboard
31 deepseek-r1-0528-qwen3-8b 0.49 Thinking May 2025 Official leaderboard
32 Command R (03-2024) 0.48 Non-thinking Mar 2024 Official leaderboard
33 Qwen3-4B 0.48 Non-thinking Apr 2025 Official leaderboard
34 Gemma 3 1B 0.33 Non-thinking Mar 2025 Official leaderboard
35 Gemma 3 270M 0.17 Non-thinking Aug 2025 Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Persian IFEval, Persian MT-Bench and PerMMLU. o3 leads 3 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Persian IFEval

35 models we track have a score here.

  1. 1 o3 0.93
  2. 2 GPT-5 mini ↑1 0.91
  3. 3 GPT-5 nano ↑7 0.90
  4. 4 Gemini 2.5 Pro Jun 2025 ↓2 0.89
  5. 5 gpt-oss-120b ↑6 0.89
  6. 6 Claude 3.7 Sonnet ↓2 0.88
  7. 7 Gemini 2.5 Flash Sep 2025 ↓1 0.88
  8. 8 GPT-4.1 ↓3 0.86
  9. 9 gpt-oss-20b ↑7 0.86
  10. 10 Gemini 2.0 Flash Feb 2025 ↓3 0.85

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5 mini
Thinking minimal rea…
0.72
GPT-5 nano
Thinking minimal rea…
0.66
gpt-oss-120b
Thinking low reasoni…
0.66
gpt-oss-20b
Thinking low reasoni…
0.62

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
o3 Thinking 0.75
Gemini 2.5 Pro Jun 2025 Thinking 0.73
GPT-5 mini Thinking 0.72
Claude 3.7 Sonnet Non-thinking 0.71
GPT-4.1 Non-thinking 0.70
Gemini 2.5 Flash Sep 2025 Thinking 0.70
Gemini 2.0 Flash Feb 2025 Non-thinking 0.69
GPT-4o Mar 2025 Non-thinking 0.69
Gemini 2.0 Flash-Lite Non-thinking 0.66
GPT-5 nano Thinking 0.66
gpt-oss-120b Thinking 0.66
GPT-5 mini minimal reasoning 0.66
GPT-4.1 mini Non-thinking 0.66
Gemma 3 27B Non-thinking 0.62
GPT-4o mini Non-thinking 0.62
gpt-oss-120b low reasoning 0.62
Qwen3-32B Non-thinking 0.62
gpt-oss-20b Thinking 0.62
Llama 3.3 70B Instruct Non-thinking 0.61
Cohere Command A Non-thinking 0.61
Gemma 3 12B Non-thinking 0.60
Qwen3-30B-A3B Non-thinking 0.59
Qwen3-14B Non-thinking 0.59
gpt-oss-20b low reasoning 0.58
Llama 4 Scout Non-thinking 0.57
Command R+ Non-thinking 0.57
Mistral Small 3.1 Non-thinking 0.56
Claude 3.5 Haiku Non-thinking 0.55
GPT-4.1 nano Non-thinking 0.55
Gemma 3n E4B Non-thinking 0.54
Qwen3-8B Non-thinking 0.53
GPT-5 nano minimal reasoning 0.52
Gemma 3 4B Non-thinking 0.50
Aya Expanse 32B Non-thinking 0.49
deepseek-r1-0528-qwen3-8b Thinking 0.49
Command R (03-2024) Non-thinking 0.48
Qwen3-4B Non-thinking 0.48
Gemma 3 1B Non-thinking 0.33
Gemma 3 270M Non-thinking 0.17

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-5 mini
Thinking minimal rea…
0.72
GPT-5 nano
Thinking minimal rea…
0.66
gpt-oss-120b
Thinking low reasoni…
0.66
gpt-oss-20b
Thinking low reasoni…
0.62

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
o3 Thinking 0.75
Gemini 2.5 Pro Jun 2025 Thinking 0.73
GPT-5 mini Thinking 0.72
Claude 3.7 Sonnet Non-thinking 0.71
GPT-4.1 Non-thinking 0.70
Gemini 2.5 Flash Sep 2025 Thinking 0.70
Gemini 2.0 Flash Feb 2025 Non-thinking 0.69
GPT-4o Mar 2025 Non-thinking 0.69
Gemini 2.0 Flash-Lite Non-thinking 0.66
GPT-5 nano Thinking 0.66
gpt-oss-120b Thinking 0.66
GPT-5 mini minimal reasoning 0.66
GPT-4.1 mini Non-thinking 0.66
Gemma 3 27B Non-thinking 0.62
GPT-4o mini Non-thinking 0.62
gpt-oss-120b low reasoning 0.62
Qwen3-32B Non-thinking 0.62
gpt-oss-20b Thinking 0.62
Llama 3.3 70B Instruct Non-thinking 0.61
Cohere Command A Non-thinking 0.61
Gemma 3 12B Non-thinking 0.60
Qwen3-30B-A3B Non-thinking 0.59
Qwen3-14B Non-thinking 0.59
gpt-oss-20b low reasoning 0.58
Llama 4 Scout Non-thinking 0.57
Command R+ Non-thinking 0.57
Mistral Small 3.1 Non-thinking 0.56
Claude 3.5 Haiku Non-thinking 0.55
GPT-4.1 nano Non-thinking 0.55
Gemma 3n E4B Non-thinking 0.54
Qwen3-8B Non-thinking 0.53
GPT-5 nano minimal reasoning 0.52
Gemma 3 4B Non-thinking 0.50
Aya Expanse 32B Non-thinking 0.49
deepseek-r1-0528-qwen3-8b Thinking 0.49
Command R (03-2024) Non-thinking 0.48
Qwen3-4B Non-thinking 0.48
Gemma 3 1B Non-thinking 0.33
Gemma 3 270M Non-thinking 0.17

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Not tested since Sep 2025
o3 0.75

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
0.00 0.20 0.40 0.60 0.80 2024 2025 0.00 0.20 0.40 0.60 0.80 2024 2025

Source: Official MIZAN: A Persian LLM Leaderboard leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Multilingual

All 545 in Multilingual
  • MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
  • WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET. K-EXAONE 90.5%
  • MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language. Qwen3.7-Max 87%
  • CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories. Solar Open 2 90.7%
  • Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
  • MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 14 months ago, in September 2025, for Gemini 2.5 Flash. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (41 entries).
  • Pass Models scored 35 models we track, plus 6 other entries on its leaderboard.
  • Warning Newest score The newest came 14 months ago, in September 2025, for Gemini 2.5 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is o3 (#84).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on mcinext-mizan-llm-leaderboard.hf.space: 39 rows for models we track and 6 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check