MiniCPM-V 4.5 — Overview

A vision-language MLLM with strong performance on image / video understanding, OCR, and document parsing. Released alongside MiniCPM-o 4.5.

Looking at this for a new project? Consider MiniCPM-V 4.6 instead — it has a longer context, faster vision tower, and is the actively-recommended release.

Highlights

• 9B parameters, Qwen3 backbone
• Image, multi-image, and video understanding (no audio)
• One model with optional thinking mode (toggle via enable_thinking)
• HuggingFace: openbmb/MiniCPM-V-4_5 · ModelScope: OpenBMB/MiniCPM-V-4_5

Where to start

• Inference: Single image, Multi-image, Video, OCR, PDF, Grounding
• Deploy: vLLM, SGLang, llama.cpp, Ollama
• Quantize: GGUF, BNB, AWQ