KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit, quantized by OptiQ for Apple Silicon; mixture-of-experts.

MoEexpert-pruned

On disk12.3 GB

Memoryabout 16 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 mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

OptiQ Code

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

optiq serve --model mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B
optiq code

OptiQ Lab

optiq lab --model mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

Claude Code

optiq run claude -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

Codex

optiq run codex -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

OpenCode

optiq run opencode -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

Mistral Vibe

optiq run vibe -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

Hermes Agent

optiq run hermes -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

OpenClaw

optiq run openclaw -m mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B

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/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

Family	Qwen · Qwen docs
Architecture	Qwen3.5 MoE
Base model	mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit
Size on disk	12.3 GB
Memory needed	about 16 GB
Images	no, text only
Downloads	894 in the last month
Updated	2026-09-15
Hugging Face	mlx-community/KAT-Coder-V2.5-Dev-OptiQ-4bit-REAP-18B