Skip to content

Benchmark

OCRBench

OCR benchmarks measure how well models read text in images and documents. Scores are percentages, except where a source reports on a different scale. Check each score's scale before comparing. Higher is better.

Models measured
64
Scores recorded
68
Sources
3
IQ weight
0.00

Score distribution

528.4%
Mean
701%
Median
358%
Std dev
0.877%
10th percentile
862%
90th percentile
0.811%
Min

This benchmark has a weight of 0.00 for chat models, measured over 64 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1InternVL3 78B· OpenVLM Leaderboard data908
2SenseNova-V6-Pro· OpenVLM Leaderboard data895
3Qwen2-VL-72B· OpenVLM Leaderboard data
Values from 2 sources
888
4Qwen2.5-VL 7B Instruct· OpenVLM Leaderboard data
Values from 2 sources
888
5internvl3-38b· OpenVLM Leaderboard data886
6internvl3-8b· OpenVLM Leaderboard data884
7Qwen2.5 VL 72B Instruct· OpenVLM Leaderboard data
Values from 2 sources
882
8Gemini 2.5 Pro· OpenVLM Leaderboard data862
9Hunyuan Standard Vision· OpenVLM Leaderboard data861
10MiniCPM-V 2.6· OpenVLM Leaderboard data852
11Gemini 2.0 Flash· OpenVLM Leaderboard data851
12GPT-4.5 Preview· OpenVLM Leaderboard data845
13Phi 4 Multimodal Instruct· OpenVLM Leaderboard data
Values from 2 sources
844
14GLM 4 Vision Plus· OpenVLM Leaderboard data843
15GPT-4.1 mini· OpenVLM Leaderboard data840
16GPT 4.1· OpenVLM Leaderboard data834
17GPT-5 mini· OpenVLM Leaderboard data828
18qwen2.5-vl-3b-instruct· OpenVLM Leaderboard data828
19InternVL2 26B· OpenVLM Leaderboard data825
20ChatGPT-4o Latest· OpenVLM Leaderboard data822
21GLM 4 Vision· OpenVLM Leaderboard data814
22DeepSeek VL2· OpenVLM Leaderboard data
Values from 2 sources
809
23GPT 5· OpenVLM Leaderboard data807
24Moonshot V1 8K· OpenVLM Leaderboard data790
25InternVL2 4B· OpenVLM Leaderboard data784
26Llama 3.2 90B Vision Instruct· OpenVLM Leaderboard data783
27gemini-1.5-pro-002· OpenVLM Leaderboard data770
28Gemini 1.5 Pro· OpenVLM Leaderboard data754
29Gemini 1.5 Flash· OpenVLM Leaderboard data753
30Gemma 3 27B IT· OpenVLM Leaderboard data753
31Llama 3.2 11B Vision Instruct· OpenVLM Leaderboard data753
32GPT-5 nano· OpenVLM Leaderboard data747
33gemini-1.5-flash-002· OpenVLM Leaderboard data733
34Qwen-VL Plus· OpenVLM Leaderboard data726
35GPT-4.1 nano· OpenVLM Leaderboard data710
36Gemma 3 12B IT· OpenVLM Leaderboard data702
37Molmo 72B· OpenVLM Leaderboard data701
38Yi Vision· OpenVLM Leaderboard data701
39Claude 3 Opus· OpenVLM Leaderboard data694
40Pixtral 12B· OpenVLM Leaderboard data685
41Qwen-VL Max· OpenVLM Leaderboard data684
42gemini-1.0-pro· OpenVLM Leaderboard data680
43Gemma 3 4B· OpenVLM Leaderboard data660
44Claude 3 Haiku· OpenVLM Leaderboard data658
45Molmo 7B-D· OpenVLM Leaderboard data656
46Claude 3 Sonnet· OpenVLM Leaderboard data646
47Phi-3.5 Vision Instruct· OpenVLM Leaderboard data599
48Moondream2· OpenVLM Leaderboard data585
49Grok 2 Vision 1212· OpenVLM Leaderboard data555
50kosmos-2· OpenVLM Leaderboard data244

Showing 1–50 of 68 results.

Data credits: OpenVLM Leaderboard (OpenCompass / VLMEvalKit), Apache-2.0 (licence) · Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.