gpt-oss-20b-optiq-3bpw

openai/gpt-oss-20b, quantized with OptiQ by mstyslavity for Apple Silicon.

community

On disk16.6 GB

Memoryabout 21 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 mstyslavity/gpt-oss-20b-optiq-3bpw

OptiQ Code

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

optiq serve --model mstyslavity/gpt-oss-20b-optiq-3bpw
optiq code

OptiQ Lab

optiq lab --model mstyslavity/gpt-oss-20b-optiq-3bpw

Claude Code

optiq run claude -m mstyslavity/gpt-oss-20b-optiq-3bpw

Codex

optiq run codex -m mstyslavity/gpt-oss-20b-optiq-3bpw

OpenCode

optiq run opencode -m mstyslavity/gpt-oss-20b-optiq-3bpw

Mistral Vibe

optiq run vibe -m mstyslavity/gpt-oss-20b-optiq-3bpw

Hermes Agent

optiq run hermes -m mstyslavity/gpt-oss-20b-optiq-3bpw

OpenClaw

optiq run openclaw -m mstyslavity/gpt-oss-20b-optiq-3bpw

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("mstyslavity/gpt-oss-20b-optiq-3bpw")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

Family	gpt-oss
Architecture	gpt-oss
Base model	openai/gpt-oss-20b
Size on disk	16.6 GB
Memory needed	about 21 GB
Images	no, text only
Downloads	231 in the last month
Updated	2026-08-21
Hugging Face	mstyslavity/gpt-oss-20b-optiq-3bpw