OpenBMB 开源 1.3B OCR 模型 MiniCPM-OCR 1.0，开放权重达到开放 SOTA
OpenBMB 放了个 1.3B 的开源 OCR 模型，分数刷到开放 SOTA，还配了个挑错的 Jev 模型，部署到 vLLM、llama.cpp 都支持。
OpenBMB 发布开源 OCR 模型 MiniCPM-OCR 1.0，仅 1.3B 参数。OCRBench v2 得分从 47.4 提升到 54.7，CHAOS-Bench 从 21.28 提升到 29.78，吞吐量达 PaddleOCR-VL-1.6 的 1.49 倍。模型上线首日即支持 vLLM、SGLang、llama.cpp 和 Ollama，并提供量化版本。同时发布 2B 质量评估模型 MiniCPM-OCR Jev，用于标记不可靠的 OCR 输出，在 JevOCRBench 上得分 93.6，超过 JEV-35B-A3B 的 90.1，接近 Qwen3.5-397B-A17B 的 93.8。成本方面，自建推理可低于每千页 1 美元，而专有 OCR 产品通常为每千页 10 美元。
OpenBMB presents MiniCPM-OCR 1.0, a 1.3B open-source OCR model with open-SOTA results across major benchmarks. - OCRBench v2: 47.4 -> 54.7 - CHAOS-Bench: 21.28 -> 29.78 - Competitive throughput: 1.49× PaddleOCR-VL-1.6
Recently, open OCR models have rapidly improved to the point where they often exceed proprietary OCR products offered by hyperscalers, while costing an order of magnitude less. This release pushes the trend further. Honestly, it makes no sense to keep paying $10 per 1k pages on proprietary products anymore when you can pay <$1 per 1k pages.
Day-one support: vLLM, SGLang, llama.cpp and Ollama, plus quantized versions.
Also released: MiniCPM-OCR Jev, a 2B quality judge for flagging unreliable OCR outputs and prioritizing review. It achieves 93.6 on JevOCRBench, outperforming JEV-35B-A3B (90.1) and approaching Qwen3.5-397B-A17B (93.8).
HF (MiniCPM-OCR 1.0): https://t.co/D0yFoe648o HF (MiniCPM-OCR Jev): https://t.co/E19N6rrtRm GitHub: https://t.co/42TSMHtk1D @OpenBMB
