None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
😢 Every free AI generation costs us real GPU money. We pay out of pocket to keep this free for you.
💎 Upgrade to Pro and unlock:
- ✓HD Download
- ✓No Watermark
- ✓Unlimited Generations
xiaomi ocr 0
Xiaomi OCR 0 is a compact 0.8B vision-language model that turns dense pages, tables, formulas and handwriting into clean, structured text.
All Tools
Discover our comprehensive AI-powered animation toolkit
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
Veo3.1
Create Stunning Videos with Veo3.1
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
Seedance 2.0
The Future of AI Video Is Here.
Why Use Xiaomi OCR 0
One 0.8B checkpoint covers document parsing, OCR-centric question answering and key information extraction, so a single model replaces a stack of narrow tools.
- Unified 0.8B ModelParsing, OCR-oriented VQA and field extraction live in one compact checkpoint.
- Top-Ranked AccuracyFirst place on Real5-OmniDocBench at 95.24, plus runner-up spots on OmniDocBench v1.6 and its Wild counterpart.
- Steady on Real ScansHolds up on scanned, warped, skewed and photographed pages, not only clean PDFs.
How to Use Xiaomi OCR 0
Three steps take a page image to clean, structured output.
Xiaomi OCR 0 Features
A small footprint, robust parsing, handwriting support and open weights — the practical strengths of this 0.8B OCR model.
Document Parsing
Handles standard pages plus scanned, warped, skewed and recaptured documents.
Tables & Formulas
Its formula score reaches 97.84 on UniMER-Test, while table structure lands at 92.27 on PubTabNet.
Handwriting & Calligraphy
Scores 73 on Chronicles-OCR mature scripts, level with Seed 2.0 Pro.
OCR-Centric VQA
Across five visual question answering suites it averages 83.2, edging past a 2B rival while using a fraction of the size.
Field Extraction
Pulls structured fields from receipts and forms instead of plain transcription.
Open & Local-Ready
Apache-2.0 weights plus GGUF and BF16 builds for llama.cpp, Ollama and Transformers.
Xiaomi OCR 0 — FAQ
Common questions about this compact OCR model, its benchmarks and local deployment.
What is this model?
A 0.8B vision-language checkpoint built specifically for OCR work — it reads documents and handles OCR-centric questions in one pass.
Which benchmarks does it top?
It takes first place on Real5-OmniDocBench and second place on both OmniDocBench v1.6 and Wild-OmniDocBench.
Can it read handwriting?
Yes. Clerical, cursive, running and regular calligraphic scripts are supported, with a score of 73 on Chronicles-OCR.
How was it trained?
Its recipe pairs roughly 170M OCR-focused samples with Q-Mask text anchoring, continued pretraining, and mixed-task reinforcement learning.
Can I run it locally?
Apache-2.0 weights are released, with community GGUF and BF16 conversions for llama.cpp, Ollama and Transformers.
Does it only transcribe text?
No. It also answers questions about a page and extracts key fields into structured output.
Try Xiaomi OCR 0 on Your Own Documents
Drop in a page image and let Xiaomi OCR 0 return parsed text, tables and structured fields in seconds.