mlx-optiq
← Models

Qwen3.6-27B-OptiQ-4bit

Qwen/Qwen3.6-27B, quantized with OptiQ by Meguriai for Apple Silicon.

community
On disk18.7 GB
Memoryabout 23 GB
Capability Score83.60

Capability Score breakdown

The six metrics behind this model's Capability Score of 83.60, as reported in the model card.

Capability Score breakdown: Qwen3.6-27B-OptiQ-4bitMMLU: This model 87.4; GSM8K: This model 92.0; IFEval: This model 74.1; BFCL-V3: This model 74.0; HumanEval: This model 90.2MMLU87.4GSM8K92.0IFEval74.1BFCL-V374.0HumanEval90.2

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 Meguriai/Qwen3.6-27B-OptiQ-4bit

OptiQ Code

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

optiq serve --model Meguriai/Qwen3.6-27B-OptiQ-4bit
optiq code

OptiQ Lab

optiq lab --model Meguriai/Qwen3.6-27B-OptiQ-4bit

Claude Code

optiq run claude -m Meguriai/Qwen3.6-27B-OptiQ-4bit

Codex

optiq run codex -m Meguriai/Qwen3.6-27B-OptiQ-4bit

OpenCode

optiq run opencode -m Meguriai/Qwen3.6-27B-OptiQ-4bit

Mistral Vibe

optiq run vibe -m Meguriai/Qwen3.6-27B-OptiQ-4bit

Hermes Agent

optiq run hermes -m Meguriai/Qwen3.6-27B-OptiQ-4bit

OpenClaw

optiq run openclaw -m Meguriai/Qwen3.6-27B-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("Meguriai/Qwen3.6-27B-OptiQ-4bit")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

FamilyQwen · Qwen docs
ArchitectureQwen3.5
Base modelQwen/Qwen3.6-27B
Size on disk18.7 GB
Memory neededabout 23 GB
Imagesno, text only
Capability Score83.60 (from the model card)
Downloads12 in the last month
Updated2026-05-25
Hugging FaceMeguriai/Qwen3.6-27B-OptiQ-4bit