Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

Kwai-Keye/Keye-VL-2.0-30B-A3B, quantized with OptiQ by lemuralabs for Apple Silicon; mixture-of-experts.

communityMoE

On disk15.0 GB

Memoryabout 19 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 lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

OptiQ Code

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

optiq serve --model lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw
optiq code

OptiQ Lab

optiq lab --model lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

Claude Code

optiq run claude -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

Codex

optiq run codex -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

OpenCode

optiq run opencode -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

Mistral Vibe

optiq run vibe -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

Hermes Agent

optiq run hermes -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

OpenClaw

optiq run openclaw -m lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw

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("lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw")
print(generate(model, tokenizer, prompt="Hello", max_tokens=256))

Details

Family	Keyevl2
Architecture	Keyevl2
Base model	Kwai-Keye/Keye-VL-2.0-30B-A3B
Size on disk	15.0 GB
Memory needed	about 19 GB
Images	no, text only
Downloads	64 in the last month
Updated	2026-08-05
Hugging Face	lemuralabs/Keye-VL-2.0-30B-A3B-uncensored-mlx-optiq-3.7bpw