MiniCPM-V-4.7-35B-A3B

What is MiniCPM-V-4.7-35B-A3B

MiniCPM-V-4.7-35B-A3B is an open-source multimodal large model released by OpenBMB. It supports multimodal understanding of images, video and more, with a 35B-A3B parameter scale.

Key features

• Supports multimodal understanding of images, video and more
• Uses a 35B-A3B MoE architecture; the first MoE model in the MiniCPM-V series
• Integrates Lookahead Expert Scheduling (LES) technology for lossless on-device smooth inference
• Uses fine-grained per-layer quantisation, enabling it to run on iPhone with 3 GB of memory
• Based on intra-ViT early compression technology that compresses visual tokens early inside the vision encoder

Who it is for

Suitable for developers, researchers and end users who need to run multimodal large models on device, especially in resource-constrained mobile scenarios.

How to get started

1. Visit the Hugging Face page https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B to download the model weights
2. Refer to the Usage section in the model card for integration and inference
3. Combine with DSpark for on-device deployment and inference

Compiled from public descriptions on the tool's official website / README / launch page. Pricing and user numbers have not been verified; please check the official website. Some sections were drafted with AI assistance. Written in Chinese and English; other languages are machine-translated.