Open-source models | Tianyu Yu

MiniCPM-OCR

Wed, 28 Oct 2026 00:00:00 +0000

The first complete OCR solution combining fast, accurate parsing with reliable OCR quality assessment.

MiniCPM-OCR 1.0

MiniCPM-OCR 1.0 (1.3B) is the first OCR-specific model in the MiniCPM-V/o family. It focuses on text-centric document understanding, covering text, tables, formulas, charts, seals, and structured information extraction.

🏆 SOTA document parsing: MiniCPM-OCR 1.0 achieves a SOTA score of 93.68 on Real5-OmniDocBench, a benchmark for real-world document parsing, and demonstrates leading parsing capabilities on OmniDocBench v1.6, Wild-OmniDocBench, and olmOCR-Bench. It supports Chinese and English documents and PDFs, handling both standard documents and challenging capture conditions, including scans, curved pages, photographs, skewed pages, and recaptured documents.
⚡ Efficient inference. With DSpark decoding acceleration (DSpark-BF16, DSpark-GGUF), MiniCPM-OCR 1.0 improves end-to-end processing efficiency by 55%. Its throughput is approximately 1.49× that of PaddleOCR-VL-1.6 and 3.62× that of MinerU2.5 Pro, reducing the time needed for large-scale document parsing.

MiniCPM-OCR Jev

MiniCPM-OCR Jev is an OCR quality decision model for text, formulas, and tables. Given an image and its corresponding OCR result, it returns a quality score from 0 to 1 to help select reliable results and locate content that needs review.

Official introduction

MiniCPM-V 4.7

Mon, 12 Oct 2026 00:00:00 +0000

MiniCPM-V 4.7 is the latest model in the MiniCPM-V series, available in 35B-A3B and 1B variants. The 35B-A3B variant is the series’ first MoE model.

🚀 A new performance–efficiency Pareto frontier. The 35B-A3B and 1B variants each advance the performance–efficiency Pareto frontier for multimodal large language models. In Instruct mode, 35B-A3B achieves a higher overall score in less inference time than Qwen3.6 35B-A3B; the 1B variant further improves on MiniCPM-V 4.6’s core capabilities and substantially outperforms Qwen3.5 0.8B in both overall score and inference efficiency.
📱 The first open-source MoE to run smoothly on-device without compromising performance. Powered by the latest Lookahead Expert Scheduling (LES) inference acceleration technique, MiniCPM-V 4.7 35B-A3B becomes the first open-source MoE model to run smoothly on-device without compromising performance, achieving a decoding speed of 13+ tokens/s with just 3 GB of memory on iPhone 17 Pro Max. We also support DSpark speculative decoding, with dedicated heads available for 1B (DSpark-BF16, DSpark-GGUF) and 35B-A3B (DSpark-BF16, DSpark-GGUF).
🛠️ Broad platform support and easy customization. MiniCPM-V 4.7 supports iOS, Android, and HarmonyOS, and supports chip platforms from Intel, AMD, Arm, Rockchip, and Houmo, covering a diverse range of mobile devices and local computing scenarios. We also open-source Self-Replay Fine-Tuning (SRFT), helping developers extend model capabilities at low cost while reducing forgetting of existing capabilities.

Official introduction

MiniCPM-V 4.6

Mon, 11 May 2026 00:00:00 +0000

The model is built based on SigLIP2-400M and the Qwen3.5-0.8B LLM. It inherits the strong single-image, multi-image, and video understanding capabilities of MiniCPM-V family, while significantly improving computation efficiency. It also introduces mixed 4x/16x visual token compression.

Ultra-Efficient Architecture. Based on the latest technique in LLaVA-UHD v4, MiniCPM-V 4.6 reduces the visual encoding computation FLOPs by more than 50%. It enables MiniCPM-V 4.6 to achieve better efficiency to even smaller models, achieving ~1.5x token throughput compared to Qwen3.5-0.8B. It also supports mixed 4x/16x visual token compression rate, allowing flexible switching between accuracy and speed.
Broad Mobile Platform Coverage. MiniCPM-V 4.6 can be deployed across all three mainstream mobile platforms — iOS, Android, and HarmonyOS.

Official introduction

MiniCPM-o 4.5

Tue, 03 Feb 2026 00:00:00 +0000

The first full-duplex omni-modal large language model.

MiniCPM-o 4.5 is the latest model in the MiniCPM-o series, a 9B end-to-end omnimodal model. It can continuously receive video and audio while generating speech and text, enabling real-time interactions that see, hear, and speak simultaneously.

Model highlights:

🔥 Vision capability approaching Gemini 2.5 Flash. It scores 77.6 on average across 8 popular OpenCompass benchmarks. With only 9B parameters, its visual understanding surpasses GPT-4o and Gemini 2.0 Pro, and a single model supports both instruct and thinking modes.
🎙 Strong speech capability. It supports bilingual real-time speech conversation with configurable voices in English and Chinese, making conversations more natural, expressive, and stable. A simple reference audio clip is enough for voice cloning and role play, and its voice cloning performance surpasses strong TTS tools such as CosyVoice2.
🎬 Full-duplex, proactive live interaction. It processes real-time, continuous video and audio input streams simultaneously while generating concurrent text and speech output streams in an end-to-end fashion, without mutual blocking. This allows MiniCPM-o 4.5 to see, hear, and speak simultaneously, enabling smooth, real-time omnimodal conversations. Beyond responding to requests, the model can also interact proactively, for example by initiating reminders or comments based on its continuous understanding of the scene.

Official introduction

Earlier MiniCPM-V/o Models

Thu, 01 Jan 2026 00:00:00 +0000

Earlier releases include MiniCPM-V 4.0, MiniCPM-o 2.6, and MiniCPM-V 2.6. See the official model archive for more releases.

I am a core contributor to the MiniCPM-V/o multimodal model family, covering post-training, evaluation, release, and iteration. I was the first student author of MiniCPM-V and the first author of MiniCPM-V 4.5.

My work focuses on post-training recipes that make efficient MLLMs strong in multimodal understanding, document intelligence, and real deployment scenarios.

See MiniCPM-V 4.7 for the latest vision-language release, and MiniCPM-OCR for document parsing and OCR quality assessment.

MiniCPM-V 4.5

Tue, 26 Aug 2025 00:00:00 +0000

The model is built on Qwen3-8B and SigLIP2-400M with a total of 8B parameters. It exhibits a significant performance improvement over previous MiniCPM-V and MiniCPM-o models, and introduces new useful features.

Efficient High-FPS and Long Video Understanding. Powered by a new unified 3D-Resampler over images and videos, MiniCPM-V 4.5 can now achieve 96x compression rate for video tokens, where 6 448x448 video frames can be jointly compressed into 64 video tokens (normally 1,536 tokens for most MLLMs). This means that the model can perceive significantly more video frames without increasing the LLM inference cost.
Controllable Hybrid Fast/Deep Thinking. MiniCPM-V 4.5 supports both fast thinking for efficient frequent usage with competitive performance, and deep thinking for more complex problem solving. To cover efficiency and performance trade-offs in different user scenarios, this fast/deep thinking mode can be switched in a highly controlled fashion.

Official introduction