Persian NLG
Models summarize Persian texts, translate between Persian and two other languages, and write questions from passages.
MIZAN ranks 45 model configurations. Its English translation tests include literary passages, subtitle dialogue, and religious texts in both directions.
- Made by
- MCINext
- Released
- May 2025
- Kind
- Translation
- Level
- University
- Scored
- Checked automatically
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Interaction
- Single turn
- Items from
- Another benchmark
- 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
Llama 3.3 70B Instruct leads with 20.1 among models released by September 2025 Llama 3.3 70B Instruct leads with 20.1 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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 20.1
Llama 3.3 70B Instruct
- 19.5 GPT-4.1
- 19.0 GPT-4.1 mini
- 19.0 GPT-4o
- 18.8
Command R+
- 18.2 GPT-5 mini
- 18.1 GPT-4o mini
- 17.8
Gemini 2.0 Flash
- 17.8 Claude 3.7 Sonnet
- 17.6 o3
- 16.8
Qwen3-32B
- 16.7 GPT-4.1 nano
- 16.6
Gemini 2.0 Flash-Lite
- 16.4 GPT-5 nano
- 16.4
Command R (03-2024)
- 16.4
Qwen3-30B-A3B
- 16.1
Qwen3-14B
- 15.7
Llama 4 Scout
- 15.6
Qwen3-8B
- 15.4 gpt-oss-120b
- 13.9
Qwen3-4B
- 13.7
Gemini 2.5 Flash
- 13.3 gpt-oss-20b
- 13.2
Mistral Small 3.1
- 12.0
Gemma 3 12B
- 12.0
Aya Expanse 32B
- 11.5
Gemini 2.5 Pro
- 11.4
Cohere Command A
- 11.3
Llama 3.2 3B
- 11.2 deepseek-r1-0528-qwen3-8b
- 10.9 Claude 3.5 Haiku
- 10.7
Gemma 3 27B
- 10.3 yandexgpt-lite
- 9.50
Gemma 3 4B
- 9.44 Mistral-7B-Instruct-v0.3
- 9.40
Gemma 3n E4B
- 9.34 deepseek-chat
- 8.81 deepseek-reasoner
- 8.23
Llama 3.2 1B
- 6.83
Gemma 3 1B
- 5.10
Gemma 3 270M
- 1
Llama 3.3 70B Instruct Non-thinking 20.1
- 2 GPT-4.1 Non-thinking 19.5
- 3 GPT-4.1 mini Non-thinking 19.0
- 4 GPT-4o Mar 2025, Non-thinking 19.0
- 5
Command R+ Non-thinking 18.8
- 6 GPT-5 mini Thinking 18.2
- 7 GPT-4o mini Non-thinking 18.1
- 8
Gemini 2.0 Flash Feb 2025, Non-thinking 17.8
- 9 Claude 3.7 Sonnet Non-thinking 17.8
- 10 o3 Thinking 17.6
- 11
Qwen3-32B Non-thinking 16.8
- 12 GPT-4.1 nano Non-thinking 16.7
- 13
Gemini 2.0 Flash-Lite Non-thinking 16.6
- 14 GPT-5 nano Thinking 16.4
- 15
Command R (03-2024) Non-thinking 16.4
- 16
Qwen3-30B-A3B Non-thinking 16.4
- 17
Qwen3-14B Non-thinking 16.1
- 18
Llama 4 Scout Non-thinking 15.7
- 19
Qwen3-8B Non-thinking 15.6
- 20 gpt-oss-120b Thinking 15.4
- 21
Qwen3-4B Non-thinking 13.9
- 22
Gemini 2.5 Flash Sep 2025, Thinking 13.7
- 23 gpt-oss-20b Thinking 13.3
- 24
Mistral Small 3.1 Non-thinking 13.2
- 25
Gemma 3 12B Non-thinking 12.0
- 26
Aya Expanse 32B Non-thinking 12.0
- 27
Gemini 2.5 Pro Jun 2025, Thinking 11.5
- 28
Cohere Command A Non-thinking 11.4
- 29
Llama 3.2 3B Non-thinking 11.3
- 30 deepseek-r1-0528-qwen3-8b Thinking 11.2
- 31 Claude 3.5 Haiku Non-thinking 10.9
- 32
Gemma 3 27B Non-thinking 10.7
- 33 yandexgpt-lite Non-thinking 10.3
- 34
Gemma 3 4B Non-thinking 9.50
- 35 Mistral-7B-Instruct-v0.3 Non-thinking 9.44
- 36
Gemma 3n E4B Non-thinking 9.40
- 37 deepseek-chat Non-thinking 9.34
- 38 deepseek-reasoner Thinking 8.81
- 39
Llama 3.2 1B Non-thinking 8.23
- 40
Gemma 3 1B Non-thinking 6.83
- 41
Gemma 3 270M Non-thinking 5.10
Source: Official Persian NLG leaderboard
- 20.1
Llama 3.3 70B Instruct
- 19.5 GPT-4.1
- 19.0 GPT-4.1 mini
- 19.0 GPT-4o
- 18.8
Command R+
- 18.2 GPT-5 mini
- 18.1 GPT-4o mini
- 17.8
Gemini 2.0 Flash
- 17.8 Claude 3.7 Sonnet
- 17.6 o3
- 16.8
Qwen3-32B
- 16.7 GPT-4.1 nano
- 16.6
Gemini 2.0 Flash-Lite
- 16.4 GPT-5 nano
- 16.4
Command R (03-2024)
- 16.4
Qwen3-30B-A3B
- 16.1
Qwen3-14B
- 15.7
Llama 4 Scout
- 15.6
Qwen3-8B
- 15.4 gpt-oss-120b
- 13.9
Qwen3-4B
- 13.7
Gemini 2.5 Flash
- 13.3 gpt-oss-20b
- 13.2
Mistral Small 3.1
- 12.0
Gemma 3 12B
- 12.0
Aya Expanse 32B
- 11.5
Gemini 2.5 Pro
- 11.4
Cohere Command A
- 11.3
Llama 3.2 3B
- 11.2 deepseek-r1-0528-qwen3-8b
- 10.9 Claude 3.5 Haiku
- 10.7
Gemma 3 27B
- 9.50
Gemma 3 4B
- 9.40
Gemma 3n E4B
- 8.23
Llama 3.2 1B
- 6.83
Gemma 3 1B
- 5.10
Gemma 3 270M
- 1
Llama 3.3 70B Instruct Non-thinking 20.1
- 2 GPT-4.1 Non-thinking 19.5
- 3 GPT-4.1 mini Non-thinking 19.0
- 4 GPT-4o Mar 2025, Non-thinking 19.0
- 5
Command R+ Non-thinking 18.8
- 6 GPT-5 mini Thinking 18.2
- 7 GPT-4o mini Non-thinking 18.1
- 8
Gemini 2.0 Flash Feb 2025, Non-thinking 17.8
- 9 Claude 3.7 Sonnet Non-thinking 17.8
- 10 o3 Thinking 17.6
- 11
Qwen3-32B Non-thinking 16.8
- 12 GPT-4.1 nano Non-thinking 16.7
- 13
Gemini 2.0 Flash-Lite Non-thinking 16.6
- 14 GPT-5 nano Thinking 16.4
- 15
Command R (03-2024) Non-thinking 16.4
- 16
Qwen3-30B-A3B Non-thinking 16.4
- 17
Qwen3-14B Non-thinking 16.1
- 18
Llama 4 Scout Non-thinking 15.7
- 19
Qwen3-8B Non-thinking 15.6
- 20 gpt-oss-120b Thinking 15.4
- 21
Qwen3-4B Non-thinking 13.9
- 22
Gemini 2.5 Flash Sep 2025, Thinking 13.7
- 23 gpt-oss-20b Thinking 13.3
- 24
Mistral Small 3.1 Non-thinking 13.2
- 25
Gemma 3 12B Non-thinking 12.0
- 26
Aya Expanse 32B Non-thinking 12.0
- 27
Gemini 2.5 Pro Jun 2025, Thinking 11.5
- 28
Cohere Command A Non-thinking 11.4
- 29
Llama 3.2 3B Non-thinking 11.3
- 30 deepseek-r1-0528-qwen3-8b Thinking 11.2
- 31 Claude 3.5 Haiku Non-thinking 10.9
- 32
Gemma 3 27B Non-thinking 10.7
- 33
Gemma 3 4B Non-thinking 9.50
- 34
Gemma 3n E4B Non-thinking 9.40
- 35
Llama 3.2 1B Non-thinking 8.23
- 36
Gemma 3 1B Non-thinking 6.83
- 37
Gemma 3 270M Non-thinking 5.10
Source: Official Persian NLG leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 20.1 | Non-thinking | Dec 2024 | Official leaderboard |
| 2 | GPT-4.1 | 19.5 | Non-thinking | Apr 2025 | Official leaderboard |
| 3 | GPT-4.1 mini | 19.0 | Non-thinking | Apr 2025 | Official leaderboard |
| 4 | GPT-4o Mar 2025 | 19.0 | Non-thinking | Mar 2025 | Official leaderboard |
| 5 | | 18.8 | Non-thinking | Aug 2024 | Official leaderboard |
| 6 | GPT-5 mini | 18.2 | Thinking | Aug 2025 | Official leaderboard |
| 7 | GPT-4o mini | 18.1 | Non-thinking | Jul 2024 | Official leaderboard |
| 8 | | 17.8 | Non-thinking | Feb 2025 | Official leaderboard |
| 9 | Claude 3.7 Sonnet | 17.8 | Non-thinking | Feb 2025 | Official leaderboard |
| 10 | o3 | 17.6 | Thinking | Apr 2025 | Official leaderboard |
| 11 | | 16.8 | Non-thinking | Apr 2025 | Official leaderboard |
| 12 | GPT-4.1 nano | 16.7 | Non-thinking | Apr 2025 | Official leaderboard |
| 13 | | 16.6 | Non-thinking | Feb 2025 | Official leaderboard |
| 14 | GPT-5 nano | 16.4 | Thinking | Aug 2025 | Official leaderboard |
| 15 | | 16.4 | Non-thinking | Mar 2024 | Official leaderboard |
| 16 | | 16.4 | Non-thinking | Apr 2025 | Official leaderboard |
| 17 | | 16.1 | Non-thinking | Apr 2025 | Official leaderboard |
| 18 | | 15.7 | Non-thinking | Apr 2025 | Official leaderboard |
| 19 | | 15.6 | Non-thinking | Apr 2025 | Official leaderboard |
| 20 | gpt-oss-120b | 15.4 | Thinking | Aug 2025 | Official leaderboard |
| 21 | | 13.9 | Non-thinking | Apr 2025 | Official leaderboard |
| 22 | | 13.7 | Thinking | Sep 2025 | Official leaderboard |
| 23 | gpt-oss-20b | 13.3 | Thinking | Aug 2025 | Official leaderboard |
| 24 | | 13.2 | Non-thinking | Mar 2025 | Official leaderboard |
| 25 | | 12.0 | Non-thinking | Mar 2025 | Official leaderboard |
| 26 | | 12.0 | Non-thinking | Oct 2024 | Official leaderboard |
| 27 | | 11.5 | Thinking | Jun 2025 | Official leaderboard |
| 28 | | 11.4 | Non-thinking | Mar 2025 | Official leaderboard |
| 29 | | 11.3 | Non-thinking | Sep 2024 | Official leaderboard |
| 30 | deepseek-r1-0528-qwen3-8b | 11.2 | Thinking | May 2025 | Official leaderboard |
| 31 | Claude 3.5 Haiku | 10.9 | Non-thinking | Oct 2024 | Official leaderboard |
| 32 | | 10.7 | Non-thinking | Mar 2025 | Official leaderboard |
| 33 | yandexgpt-lite open weights, 8B | 10.3 | Non-thinking | Official leaderboard | |
| 34 | | 9.50 | Non-thinking | Mar 2025 | Official leaderboard |
| 35 | Mistral-7B-Instruct-v0.3 open weights, 7.3B | 9.44 | Non-thinking | Official leaderboard | |
| 36 | | 9.40 | Non-thinking | Jun 2025 | Official leaderboard |
| 37 | deepseek-chat open weights, 671B | 9.34 | Non-thinking | Official leaderboard | |
| 38 | deepseek-reasoner open weights, 671B | 8.81 | Thinking | Official leaderboard | |
| 39 | | 8.23 | Non-thinking | Sep 2024 | Official leaderboard |
| 40 | | 6.83 | Non-thinking | Mar 2025 | Official leaderboard |
| 41 | | 5.10 | Non-thinking | Aug 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 20.1 | Non-thinking | Dec 2024 | Official leaderboard |
| 2 | GPT-4.1 | 19.5 | Non-thinking | Apr 2025 | Official leaderboard |
| 3 | GPT-4.1 mini | 19.0 | Non-thinking | Apr 2025 | Official leaderboard |
| 4 | GPT-4o Mar 2025 | 19.0 | Non-thinking | Mar 2025 | Official leaderboard |
| 5 | | 18.8 | Non-thinking | Aug 2024 | Official leaderboard |
| 6 | GPT-5 mini | 18.2 | Thinking | Aug 2025 | Official leaderboard |
| 7 | GPT-4o mini | 18.1 | Non-thinking | Jul 2024 | Official leaderboard |
| 8 | | 17.8 | Non-thinking | Feb 2025 | Official leaderboard |
| 9 | Claude 3.7 Sonnet | 17.8 | Non-thinking | Feb 2025 | Official leaderboard |
| 10 | o3 | 17.6 | Thinking | Apr 2025 | Official leaderboard |
| 11 | | 16.8 | Non-thinking | Apr 2025 | Official leaderboard |
| 12 | GPT-4.1 nano | 16.7 | Non-thinking | Apr 2025 | Official leaderboard |
| 13 | | 16.6 | Non-thinking | Feb 2025 | Official leaderboard |
| 14 | GPT-5 nano | 16.4 | Thinking | Aug 2025 | Official leaderboard |
| 15 | | 16.4 | Non-thinking | Mar 2024 | Official leaderboard |
| 16 | | 16.4 | Non-thinking | Apr 2025 | Official leaderboard |
| 17 | | 16.1 | Non-thinking | Apr 2025 | Official leaderboard |
| 18 | | 15.7 | Non-thinking | Apr 2025 | Official leaderboard |
| 19 | | 15.6 | Non-thinking | Apr 2025 | Official leaderboard |
| 20 | gpt-oss-120b | 15.4 | Thinking | Aug 2025 | Official leaderboard |
| 21 | | 13.9 | Non-thinking | Apr 2025 | Official leaderboard |
| 22 | | 13.7 | Thinking | Sep 2025 | Official leaderboard |
| 23 | gpt-oss-20b | 13.3 | Thinking | Aug 2025 | Official leaderboard |
| 24 | | 13.2 | Non-thinking | Mar 2025 | Official leaderboard |
| 25 | | 12.0 | Non-thinking | Mar 2025 | Official leaderboard |
| 26 | | 12.0 | Non-thinking | Oct 2024 | Official leaderboard |
| 27 | | 11.5 | Thinking | Jun 2025 | Official leaderboard |
| 28 | | 11.4 | Non-thinking | Mar 2025 | Official leaderboard |
| 29 | | 11.3 | Non-thinking | Sep 2024 | Official leaderboard |
| 30 | deepseek-r1-0528-qwen3-8b | 11.2 | Thinking | May 2025 | Official leaderboard |
| 31 | Claude 3.5 Haiku | 10.9 | Non-thinking | Oct 2024 | Official leaderboard |
| 32 | | 10.7 | Non-thinking | Mar 2025 | Official leaderboard |
| 33 | | 9.50 | Non-thinking | Mar 2025 | Official leaderboard |
| 34 | | 9.40 | Non-thinking | Jun 2025 | Official leaderboard |
| 35 | | 8.23 | Non-thinking | Sep 2024 | Official leaderboard |
| 36 | | 6.83 | Non-thinking | Mar 2025 | Official leaderboard |
| 37 | | 5.10 | Non-thinking | Aug 2025 | Official leaderboard |
Rankings by part
We read 31 parts from the benchmark’s own leaderboard in 7 groups: question-generation persianqa, translation-en2fa en2fa, summarization samsum-fa, translation-fa2en fa2en, translation-ar2fa ar2fa, summarization pnsummary and translation-fa2ar fa2ar. Each ranks models on one part of the test. Llama 3.3 70B Instruct leads 11 of them, and 8 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
rougeL precision
37 models we track have a score here.
- 1
Gemini 2.0 Flash Feb 2025 ↑7 31.4
- 2
Gemini 2.5 Flash Sep 2025 ↑20 31.2
- 3
Gemini 2.5 Pro Jun 2025 ↑24 29.8
- 4
Gemini 2.0 Flash-Lite ↑9 27.8
- 5 GPT-5 mini ↑1 26.0
- 6 GPT-4.1 ↓4 25.8
- 7
Gemma 3 12B ↑18 25.7
- 8
Llama 3.3 70B Instruct ↓7 24.5
- 9 GPT-4.1 mini ↓6 24.4
- 10
Qwen3-32B ↑1 22.8
rougeL recall
37 models we track have a score here.
- 1
Gemini 2.5 Pro Jun 2025 ↑26 41.8
- 2
Gemini 2.5 Flash Sep 2025 ↑20 41.6
- 3
Gemini 2.0 Flash Feb 2025 ↑5 41.0
- 4
Cohere Command A ↑24 39.4
- 5 GPT-4.1 ↓3 39.2
- 6 Claude 3.7 Sonnet ↑3 38.9
- 7
Gemini 2.0 Flash-Lite ↑6 38.2
- 8 GPT-5 mini ↓2 38.0
- 9
Gemma 3 12B ↑16 37.4
- 10 o3 37.3
rougeL f1 score
37 models we track have a score here.
- 1
Gemini 2.5 Flash Sep 2025 ↑21 34.5
- 2
Gemini 2.0 Flash Feb 2025 ↑6 34.4
- 3
Gemini 2.5 Pro Jun 2025 ↑24 33.2
- 4
Gemini 2.0 Flash-Lite ↑9 31.1
- 5 GPT-4.1 ↓3 29.9
- 6 GPT-5 mini 29.7
- 7
Gemma 3 12B ↑18 29.3
- 8 GPT-4.1 mini ↓5 28.2
- 9
Qwen3-32B ↑2 27.3
- 10
Llama 3.3 70B Instruct ↓9 26.9
bleu
37 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑3 21.0
- 2
Llama 3.3 70B Instruct ↓1 20.2
- 3
Command R+ ↑2 20.2
- 4 GPT-4o mini ↑3 20.1
- 5 GPT-4.1 mini ↓2 19.8
- 6 GPT-4.1 ↓4 19.7
- 7 GPT-4.1 nano ↑5 18.3
- 8 Claude 3.7 Sonnet ↑1 18.0
- 9 GPT-5 mini ↓3 17.6
- 10
Gemini 2.0 Flash-Lite ↑3 17.5
epoque bleu
37 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑3 48.1
- 2
Llama 3.3 70B Instruct ↓1 47.5
- 3 GPT-4o mini ↑4 47.1
- 4 GPT-4.1 mini ↓1 44.6
- 5
Command R+ 43.3
- 6 GPT-4.1 ↓4 41.0
- 7 GPT-4.1 nano ↑5 40.5
- 8 Claude 3.7 Sonnet ↑1 40.0
- 9
Llama 4 Scout ↑9 39.8
- 10 gpt-oss-120b ↑10 38.4
mizan bleu
37 models we track have a score here.
- 1
Command R+ ↑4 19.3
- 2 GPT-4o Mar 2025 ↑2 19.0
- 3 GPT-4.1 ↓1 19.0
- 4 GPT-4o mini ↑3 18.3
- 5 GPT-4.1 mini ↓2 17.5
- 6 Claude 3.7 Sonnet ↑3 17.0
- 7 o3 ↑3 17.0
- 8 GPT-4.1 nano ↑4 16.9
- 9 gpt-oss-120b ↑11 16.8
- 10
Llama 3.3 70B Instruct ↓9 16.6
quran bleu
37 models we track have a score here.
- 1 GPT-4.1 ↑1 16.4
- 2
Gemini 2.0 Flash Feb 2025 ↑6 16.2
- 3
Command R+ ↑2 15.3
- 4 o3 ↑6 14.5
- 5 GPT-4o Mar 2025 ↓1 14.1
- 6 GPT-4.1 nano ↑6 14.0
- 7 GPT-5 mini ↓1 13.8
- 8 GPT-4.1 mini ↓5 13.7
- 9
Gemini 2.0 Flash-Lite ↑4 13.3
- 10 Claude 3.7 Sonnet ↓1 12.3
sahife bleu
37 models we track have a score here.
- 1 GPT-4.1 ↑1 11.0
- 2
Command R+ ↑3 10.3
- 3
Gemini 2.0 Flash-Lite ↑10 10.0
- 4 o3 ↑6 9.94
- 5
Gemini 2.0 Flash Feb 2025 ↑3 9.68
- 6
Command R (03-2024) ↑9 9.22
- 7 GPT-4.1 mini ↓4 9.16
- 8 GPT-5 mini ↓2 8.96
- 9 GPT-4o Mar 2025 ↓5 8.61
- 10 Claude 3.7 Sonnet ↓1 8.33
nahj bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 5.46
- 2 GPT-4o Mar 2025 ↑2 5.29
- 3
Command R+ ↑2 5.20
- 4 GPT-4o mini ↑3 5.04
- 5
Gemini 2.0 Flash-Lite ↑8 5.01
- 6
Command R (03-2024) ↑9 4.98
- 7 GPT-4.1 mini ↓4 4.90
- 8 GPT-4.1 ↓6 4.87
- 9 o3 ↑1 4.86
- 10 GPT-4.1 nano ↑2 4.81
tep bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 7.83
- 2 o3 ↑8 7.50
- 3 GPT-4.1 ↓1 7.38
- 4 GPT-4.1 mini ↓1 7.28
- 5
Command R+ 7.08
- 6 GPT-5 nano ↑8 7.04
- 7
Command R (03-2024) ↑8 6.93
- 8 GPT-4o Mar 2025 ↓4 6.89
- 9 GPT-5 mini ↓3 6.79
- 10
Gemini 2.0 Flash-Lite ↑3 6.52
rougeL precision
37 models we track have a score here.
- 1
Command R (03-2024) ↑14 19.4
- 2
Gemini 2.0 Flash-Lite ↑11 18.1
- 3
Gemini 2.0 Flash Feb 2025 ↑5 18.1
- 4 Claude 3.5 Haiku ↑27 17.7
- 5 GPT-4.1 mini ↓2 17.1
- 6 GPT-4.1 ↓4 16.8
- 7
Gemma 3 12B ↑18 16.6
- 8 GPT-4o Mar 2025 ↓4 16.5
- 9
Gemma 3 4B ↑24 16.4
- 10
Llama 3.3 70B Instruct ↓9 16.3
rougeL recall
37 models we track have a score here.
- 1
Command R+ ↑4 41.6
- 2
Gemini 2.0 Flash Feb 2025 ↑6 41.5
- 3
Mistral Small 3.1 ↑21 40.7
- 4
Cohere Command A ↑24 40.2
- 5
Gemini 2.0 Flash-Lite ↑8 40.2
- 6 GPT-4o Mar 2025 ↓2 39.8
- 7
Aya Expanse 32B ↑19 39.8
- 8
Qwen3-4B ↑13 39.8
- 9
Gemini 2.5 Flash Sep 2025 ↑13 39.5
- 10 GPT-4o mini ↓3 39.0
rougeL f1 score
37 models we track have a score here.
- 1
Command R (03-2024) ↑14 24.3
- 2
Gemini 2.0 Flash Feb 2025 ↑6 24.1
- 3
Gemini 2.0 Flash-Lite ↑10 23.8
- 4 GPT-4.1 mini ↓1 22.5
- 5 GPT-4o Mar 2025 ↓1 22.4
- 6 Claude 3.5 Haiku ↑25 22.3
- 7
Command R+ ↓2 22.1
- 8 GPT-4.1 ↓6 21.9
- 9
Gemma 3 12B ↑16 21.9
- 10
Llama 3.3 70B Instruct ↓9 21.6
bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 25.6
- 2 GPT-5 mini ↑4 24.4
- 3 GPT-4.1 mini 23.9
- 4 o3 ↑6 23.7
- 5 GPT-4o Mar 2025 ↓1 23.4
- 6
Command R+ ↓1 23.4
- 7 GPT-4o mini 23.3
- 8 GPT-4.1 ↓6 23.1
- 9 GPT-5 nano ↑5 22.9
- 10 Claude 3.7 Sonnet ↓1 22.5
tep bleu
37 models we track have a score here.
- 1 o3 ↑9 18.6
- 2
Llama 3.3 70B Instruct ↓1 16.9
- 3 GPT-5 mini ↑3 16.6
- 4 GPT-4o Mar 2025 16.0
- 5 GPT-4.1 ↓3 15.3
- 6 GPT-4o mini ↑1 15.0
- 7 GPT-4.1 mini ↓4 14.3
- 8
Command R+ ↓3 13.9
- 9 GPT-4.1 nano ↑3 13.5
- 10 GPT-5 nano ↑4 13.5
mizan bleu
37 models we track have a score here.
- 1
Command R+ ↑4 21.3
- 2
Llama 3.3 70B Instruct ↓1 21.1
- 3 GPT-4.1 mini 20.6
- 4 GPT-5 mini ↑2 20.6
- 5 o3 ↑5 20.1
- 6 GPT-4o mini ↑1 19.7
- 7 GPT-4o Mar 2025 ↓3 19.5
- 8 GPT-4.1 ↓6 19.3
- 9 Claude 3.7 Sonnet 19.1
- 10 GPT-5 nano ↑4 18.8
quran bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 20.1
- 2 GPT-4.1 mini ↑1 17.8
- 3 Claude 3.7 Sonnet ↑6 17.4
- 4
Command R (03-2024) ↑11 17.4
- 5 o3 ↑5 17.3
- 6 GPT-5 mini 17.3
- 7 GPT-4o mini 17.3
- 8
Qwen3-30B-A3B ↑8 17.2
- 9
Command R+ ↓4 17.0
- 10 GPT-5 nano ↑4 16.4
epoque bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 51.0
- 2 GPT-5 mini ↑4 49.6
- 3 GPT-4.1 mini 48.4
- 4 GPT-5 nano ↑10 48.2
- 5
Command R+ 47.8
- 6 GPT-4o Mar 2025 ↓2 47.5
- 7 GPT-4o mini 46.8
- 8 GPT-4.1 ↓6 46.8
- 9
Qwen3-32B ↑2 45.7
- 10 Claude 3.7 Sonnet ↓1 45.4
nahj bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 9.84
- 2 GPT-4.1 nano ↑10 9.43
- 3 GPT-4o mini ↑4 9.05
- 4
Qwen3-32B ↑7 9.02
- 5 o3 ↑5 8.93
- 6 GPT-4.1 mini ↓3 8.86
- 7 GPT-5 mini ↓1 8.83
- 8 Claude 3.7 Sonnet ↑1 8.77
- 9
Qwen3-8B ↑10 8.58
- 10 GPT-5 nano ↑4 8.58
sahife bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 11.3
- 2 GPT-4.1 mini ↑1 10.4
- 3 o3 ↑7 10.4
- 4
Qwen3-30B-A3B ↑12 10.3
- 5 GPT-4.1 ↓3 10.1
- 6 GPT-4o mini ↑1 10.1
- 7 GPT-5 mini ↓1 10.1
- 8 Claude 3.7 Sonnet ↑1 9.76
- 9 GPT-4o Mar 2025 ↓5 9.53
- 10
Qwen3-32B ↑1 9.45
bleu
37 models we track have a score here.
- 1
Gemini 2.0 Flash Feb 2025 ↑7 15.7
- 2 GPT-4o Mar 2025 ↑2 15.4
- 3 GPT-4.1 ↓1 15.3
- 4 Claude 3.7 Sonnet ↑5 14.7
- 5 GPT-4.1 mini ↓2 14.3
- 6
Gemini 2.0 Flash-Lite ↑7 14.3
- 7
Command R+ ↓2 14.1
- 8 GPT-4o mini ↓1 14.0
- 9 o3 ↑1 13.7
- 10 GPT-5 mini ↓4 13.3
sahife bleu
37 models we track have a score here.
- 1
Command R+ ↑4 13.6
- 2
Gemini 2.0 Flash-Lite ↑11 13.3
- 3 Claude 3.7 Sonnet ↑6 12.9
- 4 GPT-4o Mar 2025 12.8
- 5 GPT-4o mini ↑2 12.7
- 6 GPT-4.1 ↓4 12.5
- 7 GPT-4.1 mini ↓4 12.2
- 8
Command R (03-2024) ↑7 11.9
- 9 o3 ↑1 11.5
- 10 GPT-5 mini ↓4 11.4
nahj bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 7.42
- 2 GPT-4.1 mini ↑1 6.95
- 3 GPT-4o Mar 2025 ↑1 6.60
- 4 GPT-5 mini ↑2 6.59
- 5 GPT-4o mini ↑2 6.58
- 6
Command R+ ↓1 6.53
- 7
Command R (03-2024) ↑8 6.48
- 8 Claude 3.7 Sonnet ↑1 6.43
- 9 GPT-4.1 nano ↑3 6.42
- 10 o3 6.36
quran bleu
37 models we track have a score here.
- 1
Gemini 2.0 Flash Feb 2025 ↑7 29.0
- 2 GPT-4.1 26.8
- 3 GPT-4o Mar 2025 ↑1 26.4
- 4 Claude 3.7 Sonnet ↑5 24.4
- 5
Gemini 2.0 Flash-Lite ↑8 23.9
- 6 GPT-4.1 mini ↓3 23.5
- 7 o3 ↑3 22.9
- 8 GPT-4o mini ↓1 22.2
- 9 GPT-4.1 nano ↑3 22.1
- 10 GPT-5 mini ↓4 21.9
rougeL precision
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 22.4
- 2
Llama 3.2 1B ↑33 20.6
- 3
Llama 3.2 3B ↑26 15.1
- 4
Llama 4 Scout ↑14 13.8
- 5 Claude 3.5 Haiku ↑26 13.3
- 6
Command R (03-2024) ↑9 13.0
- 7 GPT-4.1 ↓5 12.3
- 8
Gemini 2.0 Flash Feb 2025 12.3
- 9
Command R+ ↓4 12.3
- 10
Qwen3-30B-A3B ↑6 12.1
rougeL recall
37 models we track have a score here.
- 1
Gemini 2.5 Flash Sep 2025 ↑21 43.6
- 2
Mistral Small 3.1 ↑22 43.5
- 3
Gemma 3 270M ↑34 43.3
- 4
Command R+ ↑1 43.2
- 5
Cohere Command A ↑23 42.1
- 6
Qwen3-4B ↑15 41.1
- 7 GPT-4.1 mini ↓4 40.3
- 8
Qwen3-14B ↑9 40.3
- 9 deepseek-r1-0528-qwen3-8b ↑21 40.3
- 10 GPT-4o mini ↓3 39.8
rougeL f1 score
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 24.8
- 2
Llama 3.2 1B ↑33 21.9
- 3
Llama 3.2 3B ↑26 20.3
- 4
Llama 4 Scout ↑14 19.3
- 5 Claude 3.5 Haiku ↑26 18.9
- 6
Command R+ ↓1 18.6
- 7
Command R (03-2024) ↑8 18.4
- 8 GPT-4.1 ↓6 18.1
- 9
Qwen3-30B-A3B ↑7 18.0
- 10
Gemini 2.0 Flash Feb 2025 ↓2 17.9
bleu
37 models we track have a score here.
- 1
Command R+ ↑4 9.23
- 2
Llama 3.3 70B Instruct ↓1 8.60
- 3
Command R (03-2024) ↑12 8.53
- 4 o3 ↑6 8.49
- 5 GPT-4.1 ↓3 8.28
- 6
Gemini 2.0 Flash Feb 2025 ↑2 7.93
- 7 Claude 3.7 Sonnet ↑2 7.61
- 8 GPT-5 mini ↓2 7.21
- 9 GPT-4.1 mini ↓6 6.90
- 10 GPT-4o Mar 2025 ↓6 6.57
nahj bleu
37 models we track have a score here.
- 1
Command R+ ↑4 5.11
- 2
Command R (03-2024) ↑13 4.64
- 3
Llama 3.3 70B Instruct ↓2 4.41
- 4 GPT-4.1 mini ↓1 3.97
- 5 GPT-4.1 ↓3 3.84
- 6 o3 ↑4 3.79
- 7 GPT-5 mini ↓1 3.74
- 8 GPT-4o mini ↓1 3.73
- 9
Qwen3-14B ↑8 3.70
- 10 GPT-4o Mar 2025 ↓6 3.47
sahife bleu
37 models we track have a score here.
- 1
Llama 3.3 70B Instruct 8.34
- 2 o3 ↑8 8.11
- 3 GPT-4.1 ↓1 7.81
- 4 GPT-4.1 mini ↓1 7.51
- 5 GPT-4o Mar 2025 ↓1 7.32
- 6
Command R (03-2024) ↑9 7.13
- 7 GPT-4o mini 6.89
- 8 GPT-5 mini ↓2 6.36
- 9 Claude 3.7 Sonnet 6.14
- 10
Command R+ ↓5 5.90
quran bleu
37 models we track have a score here.
- 1
Command R+ ↑4 16.7
- 2
Gemini 2.0 Flash Feb 2025 ↑6 14.7
- 3
Command R (03-2024) ↑12 13.8
- 4 o3 ↑6 13.6
- 5 Claude 3.7 Sonnet ↑4 13.5
- 6 GPT-4.1 ↓4 13.2
- 7
Llama 3.3 70B Instruct ↓6 13.1
- 8 GPT-5 mini ↓2 11.5
- 9 GPT-4.1 mini ↓6 9.22
- 10 GPT-4o mini ↓3 8.96
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.
Source: Official Persian NLG leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Non-thinking | 20.1 | ||||
| GPT-4.1 | Non-thinking | 19.5 | |||
| GPT-4.1 mini | Non-thinking | 19.0 | |||
| GPT-4o Mar 2025 | Non-thinking | 19.0 | |||
| Non-thinking | 18.8 | ||||
| GPT-5 mini | Thinking | 18.2 | |||
| GPT-4o mini | Non-thinking | 18.1 | |||
| Non-thinking | 17.8 | ||||
| Claude 3.7 Sonnet | Non-thinking | 17.8 | |||
| o3 | Thinking | 17.6 | |||
| Non-thinking | 16.8 | ||||
| GPT-4.1 nano | Non-thinking | 16.7 | |||
| Non-thinking | 16.6 | ||||
| GPT-5 nano | Thinking | 16.4 | |||
| Non-thinking | 16.4 | ||||
| Non-thinking | 16.4 | ||||
| GPT-5 mini | minimal reasoning | 16.3 | |||
| Non-thinking | 16.1 | ||||
| Non-thinking | 15.7 | ||||
| Non-thinking | 15.6 | ||||
| gpt-oss-120b | Thinking | 15.4 | |||
| gpt-oss-120b | low reasoning | 14.3 | |||
| GPT-5 nano | minimal reasoning | 14.2 | |||
| Non-thinking | 13.9 | ||||
| Thinking | 13.7 | ||||
| gpt-oss-20b | Thinking | 13.3 | |||
| gpt-oss-20b | low reasoning | 13.2 | |||
| Non-thinking | 13.2 | ||||
| Non-thinking | 12.0 | ||||
| Non-thinking | 12.0 | ||||
| Thinking | 11.5 | ||||
| Non-thinking | 11.4 | ||||
| Non-thinking | 11.3 | ||||
| deepseek-r1-0528-qwen3-8b | Thinking | 11.2 | |||
| Claude 3.5 Haiku | Non-thinking | 10.9 | |||
| Non-thinking | 10.7 | ||||
| Non-thinking | 9.50 | ||||
| Non-thinking | 9.40 | ||||
| Non-thinking | 8.23 | ||||
| Non-thinking | 6.83 | ||||
| Non-thinking | 5.10 |
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.
Source: Official Persian NLG leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| Non-thinking | 20.1 | ||||
| GPT-4.1 | Non-thinking | 19.5 | |||
| GPT-4.1 mini | Non-thinking | 19.0 | |||
| GPT-4o Mar 2025 | Non-thinking | 19.0 | |||
| Non-thinking | 18.8 | ||||
| GPT-5 mini | Thinking | 18.2 | |||
| GPT-4o mini | Non-thinking | 18.1 | |||
| Non-thinking | 17.8 | ||||
| Claude 3.7 Sonnet | Non-thinking | 17.8 | |||
| o3 | Thinking | 17.6 | |||
| Non-thinking | 16.8 | ||||
| GPT-4.1 nano | Non-thinking | 16.7 | |||
| Non-thinking | 16.6 | ||||
| GPT-5 nano | Thinking | 16.4 | |||
| Non-thinking | 16.4 | ||||
| Non-thinking | 16.4 | ||||
| GPT-5 mini | minimal reasoning | 16.3 | |||
| Non-thinking | 16.1 | ||||
| Non-thinking | 15.7 | ||||
| Non-thinking | 15.6 | ||||
| gpt-oss-120b | Thinking | 15.4 | |||
| gpt-oss-120b | low reasoning | 14.3 | |||
| GPT-5 nano | minimal reasoning | 14.2 | |||
| Non-thinking | 13.9 | ||||
| Thinking | 13.7 | ||||
| gpt-oss-20b | Thinking | 13.3 | |||
| gpt-oss-20b | low reasoning | 13.2 | |||
| Non-thinking | 13.2 | ||||
| Non-thinking | 12.0 | ||||
| Non-thinking | 12.0 | ||||
| Thinking | 11.5 | ||||
| Non-thinking | 11.4 | ||||
| Non-thinking | 11.3 | ||||
| deepseek-r1-0528-qwen3-8b | Thinking | 11.2 | |||
| Claude 3.5 Haiku | Non-thinking | 10.9 | |||
| Non-thinking | 10.7 | ||||
| Non-thinking | 9.50 | ||||
| Non-thinking | 9.40 | ||||
| Non-thinking | 8.23 | ||||
| Non-thinking | 6.83 | ||||
| Non-thinking | 5.10 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Persian NLG leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Persian NLG 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
- 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 37 models we track, plus 4 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: 41 rows for models we track and 4 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.