Benchmark Almanac

OmniDocBench v1.6

OXR-1.0 97.0 leads
In use
Version 1.6

The official leaderboard hasn’t scored a model released after MiniCPM-OCR-1.0 (October 2026). Labs report their own numbers for one newer model: Inkling 2 Mini at 92.8.

Models parse 1,651 PDF pages, including nested tables, dense formulas and unusual layouts, into text, tables and formulas. Scored by the v1.5 Overall formula.

Adds 296 harder pages and a new matching method, MGAM, on the prediction side. v1.7 on April 30th, 2026 added skills-based evaluation without new pages, so it stays in this record.

Released
April 2026
Kind
Documents, charts and text in images
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Documents → Text, Tables
Interaction
Single turn
Items from
Real work and use
Run by
Anyone
Strongest model tested
GPT-5.2 #51 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 19th, 2026

In the authors’ words

Document content extraction is a critical task in computer vision, underpinning the data needs of large language models (LLMs) and retrieval-augmented generation (RAG) systems. Despite recent progress, current document parsing methods have not been fairly and comprehensively evaluated due to the narrow coverage of document types and the simplified, unrealistic evaluation procedures in existing benchmarks. To address these gaps, we introduce OmniDocBench, a novel benchmark featuring high-quality annotations across nine document sources, including academic papers, textbooks, and more challenging cases such as handwritten notes and densely typeset newspapers. OmniDocBench supports flexible, multi-level evaluations–ranging from an end-to-end assessment to the task-specific and attribute–based analysis using 19 layout categories and 15 attribute labels. We conduct a thorough evaluation of both pipeline-based methods and end-to-end vision-language models, revealing their strengths and weaknesses across different document types. OmniDocBench sets a new standard for the fair, diverse, and fine-grained evaluation in document parsing. Dataset and code are available at https://github.com/opendatalab/OmniDocBench.

Abstract of OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotations (arXiv:2412.07626, CC0)
38
Entries on its own leaderboard, 10 models we track
5
Launch posts list it, from 4 labs
1
System card reports it, from 1 lab

OXR-1.0 leads with 97.0 OXR-1.0 leads with 97.0

Best score per entry from the benchmark's own leaderboard, read on November 19th, 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. This view shows only models we track, each at its best.

OXR-1.0 leads with 97.0, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 OXR-1.0 97.0
  2. 2 TeleOCR 96.9
  3. 3 OvisOCR2 96.5
  4. 4 MiniCPM-OCR-1.0 96.4
  5. 5 PaddleOCR-VL-1.6 96.3
  6. 6 MinerU2.5-Pro 95.8
  7. 7 GLM-OCR 95.2
  8. 8 PaddleOCR-VL-1.5 94.9
  9. 9 PaddleOCR-VL 94.2
  10. 10 Unlimited-OCR 94.0
  11. 11 Qianfan-OCR 93.9
  12. 12 Youtu-Parsing 93.7
  13. 13 Ovis2.6-30B-A3B 93.7
  14. 14 Logics-Parsing-v2 93.3
  15. 15 ABot-OCR 93.3
  16. 16 FireRed-OCR 93.3
  17. 17 MinerU-2.5 93.0
  18. 18 Gemini 3 Pro 92.9
  19. 19 Gemini 3 Flash 92.6
  20. 20 MonkeyOCR V2 91.8
  21. 21 dots.ocr 90.8
  22. 22 OpenDoc-0.1B 90.7
  23. 23 DeepSeek-OCR 2 90.3
  24. 24 HunyuanOCR 90.0
  25. 25 Qwen3-VL-235B-A22B-Instruct 89.8
  26. 26 Dolphin-v2 89.5
  27. 27 OCRVerse 88.6
  28. 28 MonkeyOCR-pro-3B 88.6
  29. 29 GPT-5.2 86.6
  30. 30 Dolphin-1.5 86.5
  31. 31 MinerU-Pipeline 86.5
  32. 32 olmOCR 85.7
  33. 33 Mistral OCR 85.7
  34. 34 Kimi K2.5 84.5
  35. 35 InternVL3.5-241B-A28B 83.8
  36. 36 Nanonets-OCR-s 83.6
  37. 37 POINTS-Reader 83.4
  38. 38 Marker 78.4

Source: Official OmniDocBench v1.6 leaderboard

#EntryScoreSettingDateSource
1 OXR-1.0 97.0 Official leaderboard
2 TeleOCR 96.9 Official leaderboard
3 OvisOCR2 96.5 Official leaderboard
4 MiniCPM-OCR-1.0 96.4 Oct 2026 Official leaderboard
5 PaddleOCR-VL-1.6 96.3 Official leaderboard
6 MinerU2.5-Pro 95.8 Official leaderboard
7 GLM-OCR 95.2 Jan 2026 Official leaderboard
8 PaddleOCR-VL-1.5 94.9 Official leaderboard
9 PaddleOCR-VL 94.2 Official leaderboard
10 Unlimited-OCR 94.0 Official leaderboard
11 Qianfan-OCR 93.9 Official leaderboard
12 Youtu-Parsing 93.7 Official leaderboard
13 Ovis2.6-30B-A3B 93.7 Official leaderboard
14 Logics-Parsing-v2 93.3 Official leaderboard
15 ABot-OCR 93.3 Official leaderboard
16 FireRed-OCR 93.3 Official leaderboard
17 MinerU-2.5 93.0 Official leaderboard
18 Gemini 3 Pro 92.9 Nov 2025 Official leaderboard
19 Gemini 3 Flash 92.6 Dec 2025 Official leaderboard
20 MonkeyOCR V2 91.8 Official leaderboard
21 dots.ocr 90.8 Official leaderboard
22 OpenDoc-0.1B 90.7 Official leaderboard
23 DeepSeek-OCR 2 90.3 Jan 2026 Official leaderboard
24 HunyuanOCR 90.0 Official leaderboard
25 Qwen3-VL-235B-A22B-Instruct 89.8 Sep 2025 Official leaderboard
26 Dolphin-v2 89.5 Official leaderboard
27 OCRVerse 88.6 Official leaderboard
28 MonkeyOCR-pro-3B 88.6 Official leaderboard
29 GPT-5.2 86.6 Dec 2025 Official leaderboard
30 Dolphin-1.5 86.5 Official leaderboard
31 MinerU-Pipeline 86.5 Official leaderboard
32 olmOCR 85.7 Official leaderboard
33 Mistral OCR 85.7 Official leaderboard
34 Kimi K2.5 84.5 Jan 2026 Official leaderboard
35 InternVL3.5-241B-A28B 83.8 Aug 2025 Official leaderboard
36 Nanonets-OCR-s 83.6 Official leaderboard
37 POINTS-Reader 83.4 Official leaderboard
38 Marker 78.4 Official leaderboard
#ModelScoreSettingDateSource
1 OXR-1.0 97.0 Official leaderboard
2 MiniCPM-OCR-1.0 96.4 Oct 2026 Official leaderboard
3 GLM-OCR 95.2 Jan 2026 Official leaderboard
4 Gemini 3 Pro 92.9 Nov 2025 Official leaderboard
5 Gemini 3 Flash 92.6 Dec 2025 Official leaderboard
6 DeepSeek-OCR 2 90.3 Jan 2026 Official leaderboard
7 Qwen3-VL-235B-A22B-Instruct 89.8 Sep 2025 Official leaderboard
8 GPT-5.2 86.6 Dec 2025 Official leaderboard
9 Kimi K2.5 84.5 Jan 2026 Official leaderboard
10 InternVL3.5-241B-A28B 83.8 Aug 2025 Official leaderboard
#ModelScoreSettingDateSource
1 OXR-1.0 97.0 Official leaderboard
2 MiniCPM-OCR-1.0 96.4 Oct 2026 Official leaderboard
3 GLM-OCR 95.2 Jan 2026 Official leaderboard
4 Gemini 3 Pro 92.9 Nov 2025 Official leaderboard
5 Inkling 2 Mini 92.8 Nov 2026 Lab model card
6 Gemini 3 Flash 92.6 Dec 2025 Official leaderboard
7 DeepSeek-OCR 2 90.3 Jan 2026 Official leaderboard
8 Qwen3-VL-235B-A22B-Instruct 89.8 Sep 2025 Official leaderboard
9 GPT-5.2 86.6 Dec 2025 Official leaderboard
10 Kimi K2.5 84.5 Jan 2026 Official leaderboard
11 InternVL3.5-241B-A28B 83.8 Aug 2025 Official leaderboard

Labs report scores for 1 model in their cards. Switch the scores above to rank them together.

2 models have held number one on the official leaderboard since its release in April 2026 2 models have held number one with claims included since its release in April 2026

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

Z.ai OpenBMB (ModelBest)
Own leaderboard 2 leaders

Source: Official OmniDocBench v1.6 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 Reported by the lab in its model card
80.0 85.0 90.0 95.0 100.0 2026 2027 80.0 85.0 90.0 95.0 100.0 2026 2027

Sources: Official OmniDocBench v1.6 leaderboard and labs’ model cards

4 labs put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

SoMark
OpenBMB (ModelBest)
Baidu
Thinking Machines
DateLaunch postScoreAs the post writes it
Nov 20, 2026 Inkling 2 Mini 92.8 OmniDocBench v1.6
Oct 28, 2026 MiniCPM-OCR-1.0 96.5 OmniDocBench v1.6
Oct 27, 2026 Unlimited OCR 2 94 OmniDocBench v1.6 Overall
Oct 7, 2026 MiniCPM-V-4.7-35B-A3B 94.2 OmniDocBench_v1.6
Oct 5, 2026 OXR-1.0 97.2 OmniDocBench Overall

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Nov 20, 2026 Inkling 2 Mini model card 92.8 OmniDocBench v1.6 Capabilities

Versions

More in Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (38 entries), launch posts (5 scores) and system cards (1 score).
  • Pass Models scored 11 models we track, plus 28 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.2 (#51).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 10 rows for models we track and 28 other entries at the last read.
  • Pass Last read November 19th, 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