Xiaomi has released Xiaomi-OCR-0, an open-source AI model designed for document parsing, key-field extraction, and OCR question answering. This 0.8 billion parameter model has achieved a score of 96.83 on the OmniDocBench benchmark, ranking first among models in its size class. The model is available on Hugging Face and can be integrated with libraries like Transformers, vLLM, and SGLang for various applications. AI
IMPACT This release provides a capable, open-source option for document AI tasks, potentially accelerating adoption in specific niches.
RANK_REASON Open-source release of a new AI model by a major tech company (Xiaomi) with benchmark performance claims.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →