mlx-optiq
← Models

gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1, quantized by OptiQ for Apple Silicon; takes images.

vision
On disk9.0 GB
Memoryabout 12 GB
Parent score68.23 gemma-4-12B-it-OptiQ-4bit
Imagesyes

This model reuses the per-layer bit allocation of gemma-4-12B-it-OptiQ-4bit, whose measured Capability Score is 68.23.

The parent's Capability Score breakdown

This model has no Capability Score of its own. The chart shows the six metrics behind its parent gemma-4-12B-it-OptiQ-4bit's score of 68.23, taken from the parent's model card and compared with uniform 4-bit.

  • gemma-4-12B-it-OptiQ-4bit
  • Uniform 4-bit
The parent's Capability Score breakdown: gemma-4-12B-it-OptiQ-4bitMMLU: gemma-4-12B-it-OptiQ-4bit 42.6, Uniform 4-bit 34.4; GSM8K: gemma-4-12B-it-OptiQ-4bit 93.4, Uniform 4-bit 90.1; IFEval: gemma-4-12B-it-OptiQ-4bit 73.9, Uniform 4-bit 71.2; BFCL-V3: gemma-4-12B-it-OptiQ-4bit 71.0, Uniform 4-bit 71.5; HumanEval: gemma-4-12B-it-OptiQ-4bit 88.4, Uniform 4-bit 76.8; HashHop: gemma-4-12B-it-OptiQ-4bit 40.0, Uniform 4-bit 27.0MMLU42.634.4GSM8K93.490.1IFEval73.971.2BFCL-V371.071.5HumanEval88.476.8HashHop40.027.0

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/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

OptiQ Code

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

optiq serve --model mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit
optiq code

OptiQ Lab

optiq lab --model mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

Claude Code

optiq run claude -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

Codex

optiq run codex -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

OpenCode

optiq run opencode -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

Mistral Vibe

optiq run vibe -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

Hermes Agent

optiq run hermes -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit

OpenClaw

optiq run openclaw -m mlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-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/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

FamilyGemma-4 · Gemma-4 docs
ArchitectureGemma 4
Base modelyuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1
Size on disk9.0 GB
Memory neededabout 12 GB
Imagesyes, through optiq serve and the Lab
Family referenceThis model reuses the per-layer bit allocation of gemma-4-12B-it-OptiQ-4bit, whose measured Capability Score is 68.23.
Downloads893 in the last month
Updated2026-10-18
Hugging Facemlx-community/gemma-4-12B-coder-fable5-composer2.5-v1-OptiQ-4bit