mlx-optiq
← Models

MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit, quantized by OptiQ for Apple Silicon; mixture-of-experts, takes images.

visionMoEexpert-pruned
On disk14.6 GB
Memoryabout 18 GB
Capability Score81.91
Imagesyes

Capability Score breakdown

The six metrics behind this model's Capability Score of 81.91, as reported in the model card. Values are compared with unpruned parent.

  • This model
  • Unpruned parent
Capability Score breakdown: MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19BMMLU: This model 63.8, Unpruned parent 86.2; GSM8K: This model 90.0, Unpruned parent 94.6; IFEval: This model 79.9, Unpruned parent 85.0; BFCL-V3: This model 92.0, Unpruned parent 93.0; HumanEval: This model 87.8, Unpruned parent 90.9; HashHop: This model 78.0, Unpruned parent 76.0MMLU63.886.2GSM8K90.094.6IFEval79.985.0BFCL-V392.093.0HumanEval87.890.9HashHop78.076.0

Use it

Install

pip install mlx-optiq

Serve it

An OpenAI- and Anthropic-compatible server on this Mac. Every app below can use it.

optiq serve --model mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

OptiQ Code

The terminal coding agent uses whatever model optiq serve is running.

optiq serve --model mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B
optiq code

OptiQ Lab

optiq lab --model mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

Claude Code

optiq run claude -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

Codex

optiq run codex -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

OpenCode

optiq run opencode -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

Mistral Vibe

optiq run vibe -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

Hermes Agent

optiq run hermes -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

OpenClaw

optiq run openclaw -m mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B

optiq run starts optiq serve if it is not running, sets the agent up for this session only, and stops the server when the agent exits.

mlx-lm, text only

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

FamilyMiniCPM · MiniCPM docs
ArchitectureMiniCPM-V
Base modelmlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit
Size on disk14.6 GB
Memory neededabout 18 GB
Imagesyes, through optiq serve and the Lab
Capability Score81.91 (from the model card)
Downloads1,056 in the last month
Updated2026-10-13
Hugging Facemlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit-REAP-19B