MiniCPM-V-4.7-35B-A3B-BF16-zhenwu-FlagOS

Introduction

ModelBest, together with OpenBMB, has open-sourced the next-generation on-device multimodal large model MiniCPM-V4.7, releasing two variants at once: 1B and 35B-A3B (MoE). The Zhongzhi FlagOS community completed day-0 cross-chip adaptation covering 12 platforms — T-Head, NVIDIA, Huawei Ascend, Hygon, Kunlunxin, MetaX, Moore Threads, Tsingmicro, Iluvatar CoreX, Sunrise, Mac (M5 Pro), and CIX P1 — based on the FlagOS unified multi-chip software stack. Model images are now live on ModelScope and HuggingFace simultaneously. This adaptation extends both variants to ARM edge CPU platforms as well, enabling a viable deployment path for multimodal understanding on both data-center and end-user devices. The 35B-A3B model is the world’s first open-source MoE multimodal large model capable of lossless and smooth on-device operation, with 35B total parameters and approximately 3B activated parameters. Its core technology, Lookahead Expert Scheduling (LES), cuts on-device inference latency by preloading experts and performing computation and reading in parallel. According to official benchmarks, when paired with the on-device optimization scheme, the model achieves a generation speed of over 13 tokens per second on an iPhone 17 Pro Max. The model retains the early-compression-within-ViT technique and 4×/16× visual token compression, and supports DSpark speculative decoding to balance visual detail and inference efficiency. In official Instruct-mode evaluations, the 35B-A3B and 1B models attained higher overall scores than Qwen3.6-35B-A3B and Qwen3.5-0.8B respectively, with shorter inference time.

Integrated Deployment

• Out-of-the-box inference scripts with pre-configured hardware and software parameters
• Released FlagOS-T-Head container image supporting deployment within minutes

Consistency Validation

• Rigorously evaluated through benchmark testing: Performance and results from the FlagOS software stack are compared against native stacks on multiple public.

Evaluation Results

Benchmark Result

Operation Steps

Download FlagOS Image

docker pull harbor.baai.ac.cn/flagrelease-public/minicpmv4.7-35b-pp001-gems5.3.5-tree0.6.2-cxnone-plugin0.3.0-vllm0.24.0-cp312-pt210-hggc130-x64-2.1.1-rbd225:202610151048

Download Open-source Model Weights

pip install modelscope
modelscope download --model FlagRelease/MiniCPM-V-4.7-35B-A3B-BF16-zhenwu-FlagOS --local_dir /data/MiniCPM-V-4.7-35B-A3B

Start the Container

sudo docker run --privileged -dit \
  -e HOST_HOSTNAME=$(hostname) \
  --network=host \
  --device=/dev/infiniband \
  --ipc=host \
  --device=/dev/alixpu_ctl \
  --device=/dev/alixpu \
  --ulimit memlock=-1 \
  --ulimit stack=67108864 \
  --init \
  -v /data:/data \
  -v /mnt:/mnt \
  -w /mnt/nfs/users/jinghao \
  --name minicpmV4.7-35b \
  harbor.baai.ac.cn/flagrelease-public/minicpmv4.7-35b-pp001-gems5.3.5-tree0.6.2-cxnone-plugin0.3.0-vllm0.24.0-cp312-pt210-hggc130-x64-2.1.1-rbd225:202610151048

docker exec -it minicpmV4.7-35b /bin/bash

Start the Server

export VLLM_FL_FLAGOS_BLACKLIST=nonzero_numpy,index_put,nonzero,copy,index_put_,index,copy_,_index_put_impl_,mm
vllm serve /data/MiniCPM-V-4.7-35B-A3B \
  --trust-remote-code \
  --served-model-name MiniCPM-V-4.7-35B-A3B\
  --tensor-parallel-size 2 \
  --mm-processor-cache-gb 0 \
  --compilation-config '{"cudagraph_mode":"FULL_AND_PIECEWISE","cudagraph_capture_sizes":[1,2,4,8,16,32,64]}' \
  --host 0.0.0.0 \
  --port 8010

Service Invocation

Invocation Script

curl http://127.0.0.1:8010/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
      "messages": [{"role": "user", "content": "请给我介绍一下GPQA"}],
      "max_tokens": 128
    }'

AnythingLLM Integration Guide

1. Download & Install

• Visit the official site: https://anythingllm.com/
• Choose the appropriate version for your OS (Windows/macOS/Linux)
• Follow the installation wizard to complete the setup

2. Configuration

• Launch AnythingLLM
• Open settings (bottom left, fourth tab)
• Configure core LLM parameters
• Click "Save Settings" to apply changes

3. Model Interaction

• After model loading is complete:
• Click "New Conversation"
• Enter your question (e.g., "Explain the basics of quantum computing")
• Click the send button to get a response

Technical Overview

FlagOS is a fully open-source system software stack designed to unify the "model–system–chip" layers and foster an open, collaborative ecosystem. It enables a "develop once, run anywhere" workflow across diverse AI accelerators, unlocking hardware performance, eliminating fragmentation among vendor-specific software stacks, and substantially lowering the cost of porting and maintaining AI workloads. With core technologies such as the FlagScale, together with vllm-plugin-fl, distributed training/inference framework, FlagGems universal operator library, FlagCX communication library, and FlagTree unified compiler, the FlagRelease platform leverages the FlagOS stack to automatically produce and release various combinations of . This enables efficient and automated model migration across diverse chips, opening a new chapter for large model deployment and application.

FlagGems

FlagGems is a high-performance, generic operator libraryimplemented in Triton language. It is built on a collection of backend-neutralkernels that aims to accelerate LLM (Large-Language Models) training and inference across diverse hardware platforms.

FlagTree

FlagTree is an open source, unified compiler for multipleAI chips project dedicated to developing a diverse ecosystem of AI chip compilers and related tooling platforms, thereby fostering and strengthening the upstream and downstream Triton ecosystem. Currently in its initial phase, the project aims to maintain compatibility with existing adaptation solutions while unifying the codebase to rapidly implement single-repository multi-backend support. Forupstream model users, it provides unified compilation capabilities across multiple backends; for downstream chip manufacturers, it offers examples of Triton ecosystem integration.

FlagScale and vllm-plugin-fl

Flagscale is a comprehensive toolkit designed to supportthe entire lifecycle of large models. It builds on the strengths of several prominent open-source projects, including Megatron-LM and vLLM, to provide a robust, end-to-end solution for managing and scaling large models. vllm-plugin-fl is a vLLM plugin built on the FlagOS unified multi-chip backend, to help flagscale support multi-chip on vllm framework.

FlagCX

FlagCX is a scalable and adaptive cross-chip communication library. It serves as a platform where developers, researchers, and AI engineers can collaborate on various projects, contribute to the development of cutting-edge AI solutions, and share their work with the global community.

FlagEval Evaluation Framework

FlagEval is a comprehensive evaluation system and open platform for large models launched in 2023. It aims to establish scientific, fair, and open benchmarks, methodologies, and tools to help researchers assess model and training algorithm performance. It features: - Multi-dimensional Evaluation: Supports 800+ modelevaluations across NLP, CV, Audio, and Multimodal fields,covering 20+ downstream tasks including language understanding and image-text generation. - Industry-Grade Use Cases: Has completed horizonta1 evaluations of mainstream large models, providing authoritative benchmarks for chip-model performance validation.

Contributing

We warmly welcome global developers to join us:

1. Submit Issues to report problems
2. Create Pull Requests to contribute code
3. Improve technical documentation
4. Expand hardware adaptation support

License

The model weights are derived from OpenBMB/MiniCPM-V-4.7-35B-A3B and are open‑sourced under the Apache License 2.0: https://www.apache.org/licenses/LICENSE-2.0.txt