Benchmark Almanac

OmniParseBench

Datalab accurate 94.1 tops its leaderboard Gemini 3.8 Flash 92.9 leads the models we track Not in model cards

Document parsers convert single pages to text, and 16,288 binary unit tests on 2,937 pages check the output. The headline joins text, table and layout subscores.

The results table also lists cost per 1,000 pages and refusals. Code is Apache 2.0 and page data keeps each source's license. Datalab's own parsers lead the table.

Made by Datalab Released Oct 2026 Documents, charts and text in images All tasks public Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Short answer A number, word or short string matched to a key.
datalab-to/omni_parse_bench 175 stars Dataset Leaderboard
31
Entries on its own leaderboard, 6 models we track

Datalab accurate tops the leaderboard with 94.1

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 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 Gemini 3.8 Flash, at 92.9.

31 entries. Scroll sideways for the rest.

  1. 94.1 Datalab accurate
  2. 92.9 Gemini 3.8 Flash
  3. 92.6 Datalab balanced
  4. 91.5 GPT-6.1 Sol
  5. 90.8 Reducto
  6. 90.4 LlamaParse
  7. 90.1 chandra-ocr-2 (Datalab)
  8. 90.0 Claude Opus 5.5
  9. 89.5 Mistral OCR
  10. 88.7 Claude Sonnet 5.5
  11. 87.9 dots.mocr
  12. 87.8 GPT-5.6 Sol
  13. 87.1 MinerU2.5-Pro
  14. 87.0 Infinity-Parser2-Pro
  15. 85.9 Surya OCR 2.5 (Datalab)
  16. 85.3 OXR-1.0
  17. 83.3 Extend
  18. 83.2 MiniCPM-OCR-1.0
  19. 83.1 Surya OCR 2 (Datalab)
  20. 83.0 Azure
  21. 82.9 LightOnOCR-3-4B
  22. 81.6 LightOnOCR-3-0.8B
  23. 81.5 GLM-OCR
  24. 81.2 Unlimited-OCR 2
  25. 80.0 Nemotron 3.5 Parse
  26. 79.3 Unlimited-OCR
  27. 78.7 PaddleOCR-VL-1.6
  28. 78.4 Infinity-Parser2-Flash
  29. 75.2 DeepSeek-OCR-2
  30. 69.5 Indra OCR (Datalab)
  31. 64.6 HunyuanOCR-1.5
  1. 1 Datalab accurate 94.1
  2. 2 Gemini 3.8 Flash 92.9
  3. 3 Datalab balanced 92.6
  4. 4 GPT-6.1 Sol 91.5
  5. 5 Reducto 90.8
  6. 6 LlamaParse 90.4
  7. 7 chandra-ocr-2 (Datalab) 90.1
  8. 8 Claude Opus 5.5 90.0
  9. 9 Mistral OCR 89.5
  10. 10 Claude Sonnet 5.5 88.7
  11. 11 dots.mocr 87.9
  12. 12 GPT-5.6 Sol 87.8
  13. 13 MinerU2.5-Pro 87.1
  14. 14 Infinity-Parser2-Pro 87.0
  15. 15 Surya OCR 2.5 (Datalab) 85.9
  16. 16 OXR-1.0 85.3
  17. 17 Extend 83.3
  18. 18 MiniCPM-OCR-1.0 83.2
  19. 19 Surya OCR 2 (Datalab) 83.1
  20. 20 Azure 83.0
  21. 21 LightOnOCR-3-4B 82.9
  22. 22 LightOnOCR-3-0.8B 81.6
  23. 23 GLM-OCR 81.5
  24. 24 Unlimited-OCR 2 81.2
  25. 25 Nemotron 3.5 Parse 80.0
  26. 26 Unlimited-OCR 79.3
  27. 27 PaddleOCR-VL-1.6 78.7
  28. 28 Infinity-Parser2-Flash 78.4
  29. 29 DeepSeek-OCR-2 75.2
  30. 30 Indra OCR (Datalab) 69.5
  31. 31 HunyuanOCR-1.5 64.6
#EntryScoreSettingDateSource
1 Datalab accurate 94.1 Official leaderboard
2 Gemini 3.8 Flash 92.9 Sep 2026 Official leaderboard
3 Datalab balanced 92.6 Official leaderboard
4 GPT-6.1 Sol 91.5 Sep 2026 Official leaderboard
5 Reducto 90.8 Official leaderboard
6 LlamaParse 90.4 Official leaderboard
7 chandra-ocr-2 (Datalab) 90.1 Official leaderboard
8 Claude Opus 5.5 90.0 Sep 2026 Official leaderboard
9 Mistral OCR 89.5 Official leaderboard
10 Claude Sonnet 5.5 88.7 Sep 2026 Official leaderboard
11 dots.mocr 87.9 Official leaderboard
12 GPT-5.6 Sol 87.8 Jul 2026 Official leaderboard
13 MinerU2.5-Pro 87.1 Official leaderboard
14 Infinity-Parser2-Pro 87.0 Official leaderboard
15 Surya OCR 2.5 (Datalab) 85.9 Official leaderboard
16 OXR-1.0 85.3 Official leaderboard
17 Extend 83.3 Official leaderboard
18 MiniCPM-OCR-1.0 83.2 Official leaderboard
19 Surya OCR 2 (Datalab) 83.1 Official leaderboard
20 Azure 83.0 Official leaderboard
21 LightOnOCR-3-4B 82.9 Official leaderboard
22 LightOnOCR-3-0.8B 81.6 Official leaderboard
23 GLM-OCR 81.5 Jan 2026 Official leaderboard
24 Unlimited-OCR 2 81.2 Official leaderboard
25 Nemotron 3.5 Parse 80.0 Official leaderboard
26 Unlimited-OCR 79.3 Official leaderboard
27 PaddleOCR-VL-1.6 78.7 Official leaderboard
28 Infinity-Parser2-Flash 78.4 Official leaderboard
29 DeepSeek-OCR-2 75.2 Official leaderboard
30 Indra OCR (Datalab) 69.5 Official leaderboard
31 HunyuanOCR-1.5 64.6 Official leaderboard

Number one over time

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

Google DeepMind
Own leaderboard 1 leader
Gemini 3.8 Flash 92.9

Scores over time

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

On the benchmark’s own leaderboard
80.0 85.0 90.0 95.0 2026 80.0 85.0 90.0 95.0 2026

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Vision

All 234 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. Kimi K3.1 Pro 84.1%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%
  • DocVQA In use Models answer 50,000 questions about 12,000 scanned document images by reading their text and layout. Scoring is ANLS, a string similarity to reference answers. Seed1.5-VL 96.9%