UD Ancient Greek Perseus
Models identify dictionary forms, parts of speech, and grammatical relationships in Ancient Greek. The version 2.10 test contains 1,306 sentences from Athenaeus.
Riemenschneider and Frank report 91.14% lemma accuracy for GreTa with character embeddings. The OdyCy authors independently evaluated the same treebank.
- Made by
- Tufts University and Leipzig University
- Released
- November 2015
- Kind
- Reading in many languages
- Level
- University
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- Ancient Greek
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- New versions replace it
- Run by
- Anyone
- Based on
- A version of another benchmark
HIT-SCIR (Harbin) tops the leaderboard with 79.4%
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.
- 79.4 HIT-SCIR (Harbin)
- 74.3 Stanford (Stanford)
- 74.2 ICS PAS (Warszawa)
- 73.4 CEA LIST (Paris)
- 73.3 TurkuNLP (Turku)
- 72.5 UDPipe Future (Praha)
- 71.8 ParisNLP (Paris)
- 71.6 LATTICE (Paris)
- 70.8 AntNLP (Shanghai)
- 66.8 iParse (Pittsburgh)
- 66.2 IBM NY (Yorktown Heights)
- 66.2 NLP-Cube (București)
- 65.2 Uppsala (Uppsala)
- 65.0 SLT-Interactions (Bengaluru)
- 63.7 LeisureX (Shanghai)
- 63.3 Fudan (Shanghai)
- 58.2 UniMelb (Melbourne)
- 57.8 BASELINE UDPipe 1.2 (Praha)
- 55.3 Phoenix (Shanghai)
- 55.0 KParse (İstanbul)
- 54.6 BOUN (İstanbul)
- 52.4 CUNI x-ling (Praha)
- 46.4 ArmParser (Yerevan)
- 35.2 ONLP lab (Ra'anana)
- 1.92 HUJI (Yerushalayim)
- 0.0 SParse (İstanbul)
- 1 HIT-SCIR (Harbin) software1-P 79.4%
- 2 Stanford (Stanford) software2 74.3%
- 3 ICS PAS (Warszawa) software1-P 74.2%
- 4 CEA LIST (Paris) software1-P 73.4%
- 5 TurkuNLP (Turku) software1-P 73.3%
- 6 UDPipe Future (Praha) software1-P 72.5%
- 7 ParisNLP (Paris) software1-P 71.8%
- 8 LATTICE (Paris) software1-P 71.6%
- 9 AntNLP (Shanghai) software1-P 70.8%
- 10 iParse (Pittsburgh) software1-P 66.8%
- 11 IBM NY (Yorktown Heights) software1-P 66.2%
- 12 NLP-Cube (București) software1-P 66.2%
- 13 Uppsala (Uppsala) software1-P 65.2%
- 14 SLT-Interactions (Bengaluru) software2-P 65%
- 15 LeisureX (Shanghai) software1-P 63.7%
- 16 Fudan (Shanghai) software5-P 63.3%
- 17 UniMelb (Melbourne) software1-P 58.2%
- 18 BASELINE UDPipe 1.2 (Praha) software1-P 57.8%
- 19 Phoenix (Shanghai) software1-P 55.3%
- 20 KParse (İstanbul) software1-P 55%
- 21 BOUN (İstanbul) software2-P 54.6%
- 22 CUNI x-ling (Praha) software1-P 52.4%
- 23 ArmParser (Yerevan) software1-P 46.4%
- 24 ONLP lab (Ra'anana) software3-P 35.2%
- 25 HUJI (Yerushalayim) software1-P 1.92%
- 26 SParse (İstanbul) software1-P 0%
Source: Official UD Ancient Greek Perseus leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | HIT-SCIR (Harbin) | 79.4% | software1-P | Official leaderboard | |
| 2 | Stanford (Stanford) | 74.3% | software2 | Official leaderboard | |
| 3 | ICS PAS (Warszawa) | 74.2% | software1-P | Official leaderboard | |
| 4 | CEA LIST (Paris) | 73.4% | software1-P | Official leaderboard | |
| 5 | TurkuNLP (Turku) | 73.3% | software1-P | Official leaderboard | |
| 6 | UDPipe Future (Praha) | 72.5% | software1-P | Official leaderboard | |
| 7 | ParisNLP (Paris) | 71.8% | software1-P | Official leaderboard | |
| 8 | LATTICE (Paris) | 71.6% | software1-P | Official leaderboard | |
| 9 | AntNLP (Shanghai) | 70.8% | software1-P | Official leaderboard | |
| 10 | iParse (Pittsburgh) | 66.8% | software1-P | Official leaderboard | |
| 11 | IBM NY (Yorktown Heights) | 66.2% | software1-P | Official leaderboard | |
| 12 | NLP-Cube (București) | 66.2% | software1-P | Official leaderboard | |
| 13 | Uppsala (Uppsala) | 65.2% | software1-P | Official leaderboard | |
| 14 | SLT-Interactions (Bengaluru) | 65% | software2-P | Official leaderboard | |
| 15 | LeisureX (Shanghai) | 63.7% | software1-P | Official leaderboard | |
| 16 | Fudan (Shanghai) | 63.3% | software5-P | Official leaderboard | |
| 17 | UniMelb (Melbourne) | 58.2% | software1-P | Official leaderboard | |
| 18 | BASELINE UDPipe 1.2 (Praha) | 57.8% | software1-P | Official leaderboard | |
| 19 | Phoenix (Shanghai) | 55.3% | software1-P | Official leaderboard | |
| 20 | KParse (İstanbul) | 55% | software1-P | Official leaderboard | |
| 21 | BOUN (İstanbul) | 54.6% | software2-P | Official leaderboard | |
| 22 | CUNI x-ling (Praha) | 52.4% | software1-P | Official leaderboard | |
| 23 | ArmParser (Yerevan) | 46.4% | software1-P | Official leaderboard | |
| 24 | ONLP lab (Ra'anana) | 35.2% | software3-P | Official leaderboard | |
| 25 | HUJI (Yerushalayim) | 1.92% | software1-P | Official leaderboard | |
| 26 | SParse (İstanbul) | 0% | software1-P | Official 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
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (26 entries).
- Pass Models scored 26 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom No scores yet.
Leaderboard import 3 pass
- Pass Importer We read the board on universaldependencies.org: 26 entries at the last read, none a model we track.
- 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.