FarSick
Models rate how closely two Persian sentences agree in meaning. The original corpus also labels whether one sentence entails or contradicts the other.
MTEB scores 50 embedding models on 8,566 pairs. MIZAN scores 45 language models on 500 pairs from the original test split.
- Made by
- Zahra Ghasemi and Mohammad Ali Keyvanrad
- Released
- September 2021
- Kind
- Reading in many languages
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Prediction
- Tasks
- All public
- Measures
- Capability
- Language
- Persian
- Interaction
- Single turn
- Items from
- Translated, Another benchmark
- Refresh
- One fixed set
- Run by
- Anyone
- Last model added
- June 2025
- Scores refreshed
- November 20th, 2026
jinaai/jina-embeddings-v3 tops the leaderboard with 0.77 Qwen3-Embedding-8B leads with 0.76 among models released by June 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. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Qwen3-Embedding-8B, at 0.76. This view shows only models we track, each at its best.
The newest model with a score here came out in June 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
- 0.77 jinaai/jina-embeddings-v3
- 0.76
Qwen3-Embedding-8B
- 0.75 jinaai/jina-embeddings-v5-omni-small
- 0.75 jinaai/jina-embeddings-v5-text-small
- 0.74 codefuse-ai/F2LLM-v2-14B
- 0.74 jinaai/jina-embeddings-v5-omni-nano
- 0.74
Qwen3-Embedding-4B
- 0.74 intfloat/multilingual-e5-large-instruct
- 0.73 codefuse-ai/F2LLM-v2-8B
- 0.72 microsoft/harrier-oss-v1-0.6b
- 0.72 MCINext/Hakim-unsup
- 0.72 MCINext/Hakim
- 0.72 Reza2kn/RaahNaame-Koochik-v1
- 0.72 BAAI/bge-m3
- 0.72 Alibaba-NLP/gte-Qwen2-7B-instruct
- 0.71 codefuse-ai/F2LLM-v2-4B
- 0.71 google/embeddinggemma-300m
- 0.71 Alibaba-NLP/gte-multilingual-base
- 0.71 codefuse-ai/F2LLM-v2-1.7B
- 0.71 intfloat/multilingual-e5-large
- 0.70 microsoft/harrier-oss-v1-270m
- 0.70 intfloat/e5-mistral-7b-instruct
- 0.70 intfloat/multilingual-e5-base
- 0.70 PartAI/Tooka-SBERT
- 0.69 hotchpotch/bekko-embedding-v1-a25m
- 0.69 codefuse-ai/F2LLM-v2-0.6B
- 0.68 MCINext/Hakim-small
- 0.68 intfloat/multilingual-e5-small
- 0.67
Qwen3-Embedding-0.6B
- 0.67 BAAI/bge-m3-unsupervised
- 0.67 codefuse-ai/F2LLM-v2-330M
- 0.67 sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
- 0.66 hotchpotch/bekko-embedding-v1-a8m
- 0.66 PartAI/Tooka-SBERT-V2-Large
- 0.65 codefuse-ai/F2LLM-v2-160M
- 0.64 sentence-transformers/LaBSE
- 0.64 codefuse-ai/F2LLM-v2-80M
- 0.64 hotchpotch/bekko-static-embedding-v1
- 0.64 PartAI/Tooka-SBERT-V2-Small
- 0.62 shekar-ai/Noql
- 0.62 ibm-granite/granite-embedding-97m-multilingual-r2
- 0.61 minishlab/potion-multilingual-128M
- 0.61 openai/text-embedding-3-small
- 0.60 sentence-transformers/static-similarity-mrl-multilingual-v1
- 0.54 HooshvareLab/bert-base-parsbert-uncased
- 0.52 PartAI/TookaBERT-Base
- 0.51 m3hrdadfi/bert-zwnj-wnli-mean-tokens
- 0.50 myrkur/sentence-transformer-parsbert-fa
- 0.49 sbunlp/fabert
- 0.48 m3hrdadfi/roberta-zwnj-wnli-mean-tokens
- 1 jinaai/jina-embeddings-v3 0.77
- 2
Qwen3-Embedding-8B 0.76
- 3 jinaai/jina-embeddings-v5-omni-small 0.75
- 4 jinaai/jina-embeddings-v5-text-small 0.75
- 5 codefuse-ai/F2LLM-v2-14B 0.74
- 6 jinaai/jina-embeddings-v5-omni-nano 0.74
- 7
Qwen3-Embedding-4B 0.74
- 8 intfloat/multilingual-e5-large-instruct 0.74
- 9 codefuse-ai/F2LLM-v2-8B 0.73
- 10 microsoft/harrier-oss-v1-0.6b 0.72
- 11 MCINext/Hakim-unsup 0.72
- 12 MCINext/Hakim 0.72
- 13 Reza2kn/RaahNaame-Koochik-v1 0.72
- 14 BAAI/bge-m3 0.72
- 15 Alibaba-NLP/gte-Qwen2-7B-instruct 0.72
- 16 codefuse-ai/F2LLM-v2-4B 0.71
- 17 google/embeddinggemma-300m 0.71
- 18 Alibaba-NLP/gte-multilingual-base 0.71
- 19 codefuse-ai/F2LLM-v2-1.7B 0.71
- 20 intfloat/multilingual-e5-large 0.71
- 21 microsoft/harrier-oss-v1-270m 0.70
- 22 intfloat/e5-mistral-7b-instruct 0.70
- 23 intfloat/multilingual-e5-base 0.70
- 24 PartAI/Tooka-SBERT 0.70
- 25 hotchpotch/bekko-embedding-v1-a25m 0.69
- 26 codefuse-ai/F2LLM-v2-0.6B 0.69
- 27 MCINext/Hakim-small 0.68
- 28 intfloat/multilingual-e5-small 0.68
- 29
Qwen3-Embedding-0.6B 0.67
- 30 BAAI/bge-m3-unsupervised 0.67
- 31 codefuse-ai/F2LLM-v2-330M 0.67
- 32 sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 0.67
- 33 hotchpotch/bekko-embedding-v1-a8m 0.66
- 34 PartAI/Tooka-SBERT-V2-Large 0.66
- 35 codefuse-ai/F2LLM-v2-160M 0.65
- 36 sentence-transformers/LaBSE 0.64
- 37 codefuse-ai/F2LLM-v2-80M 0.64
- 38 hotchpotch/bekko-static-embedding-v1 0.64
- 39 PartAI/Tooka-SBERT-V2-Small 0.64
- 40 shekar-ai/Noql 0.62
- 41 ibm-granite/granite-embedding-97m-multilingual-r2 0.62
- 42 minishlab/potion-multilingual-128M 0.61
- 43 openai/text-embedding-3-small 0.61
- 44 sentence-transformers/static-similarity-mrl-multilingual-v1 0.60
- 45 HooshvareLab/bert-base-parsbert-uncased 0.54
- 46 PartAI/TookaBERT-Base 0.52
- 47 m3hrdadfi/bert-zwnj-wnli-mean-tokens 0.51
- 48 myrkur/sentence-transformer-parsbert-fa 0.50
- 49 sbunlp/fabert 0.49
- 50 m3hrdadfi/roberta-zwnj-wnli-mean-tokens 0.48
Source: Official FarSick leaderboard
- 1
Qwen3-Embedding-8B 0.76
- 2
Qwen3-Embedding-4B 0.74
- 3
Qwen3-Embedding-0.6B 0.67
Source: Official FarSick leaderboard
| # | Entry | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | jinaai/jina-embeddings-v3 open weights, 0.6B | 0.77 | huggingface.co | Official leaderboard | ||
| 2 | | 0.76 | Jun 2025 | huggingface.co | Official leaderboard | |
| 3 | jinaai/jina-embeddings-v5-omni-small open weights, 1.6B | 0.75 | huggingface.co | Official leaderboard | ||
| 4 | jinaai/jina-embeddings-v5-text-small open weights, 0.6B | 0.75 | huggingface.co | Official leaderboard | ||
| 5 | codefuse-ai/F2LLM-v2-14B open weights, 14B | 0.74 | huggingface.co | Official leaderboard | ||
| 6 | jinaai/jina-embeddings-v5-omni-nano open weights, 1B | 0.74 | huggingface.co | Official leaderboard | ||
| 7 | | 0.74 | Jun 2025 | huggingface.co | Official leaderboard | |
| 8 | intfloat/multilingual-e5-large-instruct open weights, 0.6B | 0.74 | huggingface.co | Official leaderboard | ||
| 9 | codefuse-ai/F2LLM-v2-8B open weights, 7.6B | 0.73 | huggingface.co | Official leaderboard | ||
| 10 | microsoft/harrier-oss-v1-0.6b open weights, 0.6B | 0.72 | huggingface.co | Official leaderboard | ||
| 11 | MCINext/Hakim-unsup closed weights, 0.1B | 0.72 | huggingface.co | Official leaderboard | ||
| 12 | MCINext/Hakim closed weights, 0.1B | 0.72 | huggingface.co | Official leaderboard | ||
| 13 | Reza2kn/RaahNaame-Koochik-v1 open weights, 0.3B | 0.72 | huggingface.co | Official leaderboard | ||
| 14 | BAAI/bge-m3 open weights, 0.6B | 0.72 | huggingface.co | Official leaderboard | ||
| 15 | Alibaba-NLP/gte-Qwen2-7B-instruct open weights, 7.1B | 0.72 | huggingface.co | Official leaderboard | ||
| 16 | codefuse-ai/F2LLM-v2-4B open weights, 4B | 0.71 | huggingface.co | Official leaderboard | ||
| 17 | google/embeddinggemma-300m open weights, 0.3B | 0.71 | ai.google.dev | Official leaderboard | ||
| 18 | Alibaba-NLP/gte-multilingual-base open weights, 0.3B | 0.71 | huggingface.co | Official leaderboard | ||
| 19 | codefuse-ai/F2LLM-v2-1.7B open weights, 1.7B | 0.71 | huggingface.co | Official leaderboard | ||
| 20 | intfloat/multilingual-e5-large open weights, 0.6B | 0.71 | huggingface.co | Official leaderboard | ||
| 21 | microsoft/harrier-oss-v1-270m open weights, 0.3B | 0.70 | huggingface.co | Official leaderboard | ||
| 22 | intfloat/e5-mistral-7b-instruct open weights, 7.1B | 0.70 | huggingface.co | Official leaderboard | ||
| 23 | intfloat/multilingual-e5-base open weights, 0.3B | 0.70 | huggingface.co | Official leaderboard | ||
| 24 | PartAI/Tooka-SBERT open weights, 0.4B | 0.70 | huggingface.co | Official leaderboard | ||
| 25 | hotchpotch/bekko-embedding-v1-a25m open weights, 0.1B | 0.69 | huggingface.co | Official leaderboard | ||
| 26 | codefuse-ai/F2LLM-v2-0.6B open weights, 0.6B | 0.69 | huggingface.co | Official leaderboard | ||
| 27 | MCINext/Hakim-small closed weights, 0B | 0.68 | huggingface.co | Official leaderboard | ||
| 28 | intfloat/multilingual-e5-small open weights, 0.1B | 0.68 | huggingface.co | Official leaderboard | ||
| 29 | | 0.67 | Jun 2025 | huggingface.co | Official leaderboard | |
| 30 | BAAI/bge-m3-unsupervised open weights, 0.6B | 0.67 | huggingface.co | Official leaderboard | ||
| 31 | codefuse-ai/F2LLM-v2-330M open weights, 0.3B | 0.67 | huggingface.co | Official leaderboard | ||
| 32 | sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 open weights, 0.1B | 0.67 | huggingface.co | Official leaderboard | ||
| 33 | hotchpotch/bekko-embedding-v1-a8m open weights, 0.1B | 0.66 | huggingface.co | Official leaderboard | ||
| 34 | PartAI/Tooka-SBERT-V2-Large open weights, 0.4B | 0.66 | huggingface.co | Official leaderboard | ||
| 35 | codefuse-ai/F2LLM-v2-160M open weights, 0.2B | 0.65 | huggingface.co | Official leaderboard | ||
| 36 | sentence-transformers/LaBSE open weights, 0.5B | 0.64 | huggingface.co | Official leaderboard | ||
| 37 | codefuse-ai/F2LLM-v2-80M open weights, 0.1B | 0.64 | huggingface.co | Official leaderboard | ||
| 38 | hotchpotch/bekko-static-embedding-v1 open weights, 0.2B | 0.64 | huggingface.co | Official leaderboard | ||
| 39 | PartAI/Tooka-SBERT-V2-Small open weights, 0.1B | 0.64 | huggingface.co | Official leaderboard | ||
| 40 | shekar-ai/Noql open weights, 0B | 0.62 | huggingface.co | Official leaderboard | ||
| 41 | ibm-granite/granite-embedding-97m-multilingual-r2 open weights, 0.1B | 0.62 | huggingface.co | Official leaderboard | ||
| 42 | minishlab/potion-multilingual-128M open weights, 0.1B | 0.61 | huggingface.co | Official leaderboard | ||
| 43 | openai/text-embedding-3-small closed weights | 0.61 | openai.com | Official leaderboard | ||
| 44 | sentence-transformers/static-similarity-mrl-multilingual-v1 open weights, 0.1B | 0.60 | huggingface.co | Official leaderboard | ||
| 45 | HooshvareLab/bert-base-parsbert-uncased open weights, 0.2B | 0.54 | huggingface.co | Official leaderboard | ||
| 46 | PartAI/TookaBERT-Base open weights, 0.1B | 0.52 | huggingface.co | Official leaderboard | ||
| 47 | m3hrdadfi/bert-zwnj-wnli-mean-tokens open weights, 0.1B | 0.51 | huggingface.co | Official leaderboard | ||
| 48 | myrkur/sentence-transformer-parsbert-fa open weights, 0.2B | 0.50 | huggingface.co | Official leaderboard | ||
| 49 | sbunlp/fabert open weights, 0.1B | 0.49 | huggingface.co | Official leaderboard | ||
| 50 | m3hrdadfi/roberta-zwnj-wnli-mean-tokens open weights, 0.1B | 0.48 | huggingface.co | Official leaderboard |
| # | Model | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | | 0.76 | Jun 2025 | huggingface.co | Official leaderboard | |
| 2 | | 0.74 | Jun 2025 | huggingface.co | Official leaderboard | |
| 3 | | 0.67 | Jun 2025 | huggingface.co | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official FarSick 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 FarSick 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 17 months ago, in June 2025, for Qwen3-Embedding-0.6B. 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 (50 entries).
- Pass Models scored 3 models we track, plus 47 other entries on its leaderboard.
- Warning Newest score The newest came 17 months ago, in June 2025, for Qwen3-Embedding-0.6B.
- Note Frontier models None of its models are on the capabilities index.
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on mteb-leaderboard.hf.space: 3 rows for models we track and 47 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.