all the models — AI benchmark observatory
← Benchmarks

OmniDocBench 1.5

OmniDocBench 1.5 is a comprehensive benchmark for evaluating multimodal large language models on document understanding tasks, including OCR, document parsing, information extraction, and visual question answering across diverse document types. Lower Overall Edit Distance scores are better.

id omnidocbench-1.5 · max 1 · 20 models reported

#ModelScore
1MiniMax M3
MiniMax · open
0.92
2Qwen3.7-Plus
Alibaba Cloud / Qwen Team
0.91
3Qwen3.6 Plus
Alibaba Cloud / Qwen Team
0.91
4Qwen3.8-27B
Alibaba Cloud / Qwen Team · open
0.91
5Qwen3.6-35B-A3B
Alibaba Cloud / Qwen Team · open
0.90
6Qwen3.5-122B-A10B
Alibaba Cloud / Qwen Team · open
0.90
7Qwen3.5-35B-A3B
Alibaba Cloud / Qwen Team · open
0.89
8GPT-5.4
OpenAI
0.89
9Qwen3.5-27B
Alibaba Cloud / Qwen Team · open
0.89
10Kimi K2.5
Moonshot AI · open
0.89
11GPT-5.5 Instant
OpenAI
0.88
12GPT-5.4 mini
OpenAI
0.87