BEIJING, Oct. 31, 2026 /PRNewswire/ -- ModelBest and OpenBMB jointly announced the release and open-source availability of the MiniCPM-OCR series, a family of models for efficient document parsing and parsing-quality assessment. The launch includes two models: MiniCPM-OCR 1.0 (1.3B) and MiniCPM-OCR Jev (2B).

As the first OCR-focused model in the MiniCPM family, MiniCPM-OCR 1.0 brings top-tier parsing accuracy to a compact 1.3-billion-parameter architecture optimized for edge deployment, pushing forward the balance between efficiency and performance for open-source OCR models. Built for real-world conditions, it accurately parses scanned, photographed, curved, tilted and screen-captured documents without cumbersome image pre-correction, lowering the barrier to large-scale document digitization.

The model handles dense bilingual Chinese-English content, complex mathematical formulas, and multi-level or borderless tables, consistently performing at a leading level across a broad range of mainstream industry evaluations. Beyond full-page reconstruction, it supports key information extraction (KIE), pulling specified fields from invoices, IDs and financial statements according to business instructions.

Powered by DSpark decoding acceleration, MiniCPM-OCR 1.0 significantly increases end-to-end throughput while lowering per-page latency. For enterprises processing millions of pages—such as financial contract archives, academic libraries and government records—this means processing far more documents with the same hardware and time budget, at a substantially lower computing cost.

To address the long-standing need for page-by-page manual review, ModelBest and OpenBMB also introduced MiniCPM-OCR Jev, the first OCR result decision model that acts as an "AI quality inspector." Rather than performing OCR itself, Jev compares the source image with OCR output and rapidly assigns confidence scores to text, LaTeX formulas and tables, detecting missing characters, errors and formatting misalignments. Despite its compact 2B-parameter size, Jev delivers judgment accuracy typically associated with far larger models.

Jev supports cross-model deployment and can be integrated into existing OCR workflows without replacing the underlying OCR engine. It can automatically identify reliable results for direct processing, flag potentially inaccurate outputs for human review and select the best result from multiple OCR systems. This enables enterprises to move from full manual inspection toward risk-based, automated quality control.

The MiniCPM-OCR series is available on GitHub, Hugging Face and ModelScope, and supports vLLM, SGLang, llama.cpp and Ollama, along with SDKs, APIs, Agent Skill plugins and open-source fine-tuning code. Day-0 hardware adaptation covers AMD, Arm, Houmo AI, Intel and NVIDIA. Through the FlagOS software stack, it also has Day-0 support for AI and edge chips from CIX Technology, Huawei Ascend, Hygon, Iluvatar CoreX, Kunlunxin, MetaX, Moore Threads, Sunrise, T-Head, TsingMicro, NVIDIA and Apple Mac (Arm).