mlx-optiq
← Models

MiniCPM5-2B-OptiQ-4bit

openbmb/MiniCPM5-2B, quantized by OptiQ for Apple Silicon.

On disk1.9 GB
Memoryabout 4 GB
Capability Score69.36

Capability Score breakdown

The six metrics behind this model's Capability Score of 69.36, as reported in the model card.

Capability Score breakdown: MiniCPM5-2B-OptiQ-4bitMMLU: This model 59.8; GSM8K: This model 82.1; IFEval: This model 86.7; BFCL-V3: This model 82.5; HumanEval: This model 81.1; HashHop: This model 24.0MMLU59.8GSM8K82.1IFEval86.7BFCL-V382.5HumanEval81.1HashHop24.0

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/MiniCPM5-2B-OptiQ-4bit

OptiQ Code

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

optiq serve --model mlx-community/MiniCPM5-2B-OptiQ-4bit
optiq code

OptiQ Lab

optiq lab --model mlx-community/MiniCPM5-2B-OptiQ-4bit

Claude Code

optiq run claude -m mlx-community/MiniCPM5-2B-OptiQ-4bit

Codex

optiq run codex -m mlx-community/MiniCPM5-2B-OptiQ-4bit

OpenCode

optiq run opencode -m mlx-community/MiniCPM5-2B-OptiQ-4bit

Mistral Vibe

optiq run vibe -m mlx-community/MiniCPM5-2B-OptiQ-4bit

Hermes Agent

optiq run hermes -m mlx-community/MiniCPM5-2B-OptiQ-4bit

OpenClaw

optiq run openclaw -m mlx-community/MiniCPM5-2B-OptiQ-4bit

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/MiniCPM5-2B-OptiQ-4bit")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

FamilyMiniCPM · MiniCPM docs
ArchitectureLlama
Base modelopenbmb/MiniCPM5-2B
Size on disk1.9 GB
Memory neededabout 4 GB
Imagesno, text only
Capability Score69.36 (from the model card)
Downloads1,086 in the last month
Updated2026-09-14
Hugging Facemlx-community/MiniCPM5-2B-OptiQ-4bit