CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

google/gemma-4-26b-a4b, quantized with OptiQ by deer-sec for Apple Silicon.

community

On disk28.4 GB

Memoryabout 34 GB

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 deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

OptiQ Code

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

optiq serve --model deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX
optiq code

OptiQ Lab

optiq lab --model deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

Claude Code

optiq run claude -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

Codex

optiq run codex -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

OpenCode

optiq run opencode -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

Mistral Vibe

optiq run vibe -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

Hermes Agent

optiq run hermes -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

OpenClaw

optiq run openclaw -m deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX

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("deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

Family	Gemma-4 · Gemma-4 docs
Architecture	Gemma 4
Base model	google/gemma-4-26b-a4b
Size on disk	28.4 GB
Memory needed	about 34 GB
Images	no, text only
Downloads	478 in the last month
Updated	2026-08-16
Hugging Face	deer-sec/CyberStag-Security-26B-A4B-V1-OptiQ8-MLX