Image-Text-to-Text
MLX
Safetensors
gemma4
apple-silicon
4bit
Mixture of Experts
mixture-of-experts
on-device
conversational
4-bit precision
How to use from
Hermes AgentConfigure Hermes
# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default LetheanNetwork/lemmy-mlxRun Hermes
hermesQuick Links
LetheanNetwork/lemmy-mlx
Gemma 4 26B A4B MoE in MLX format, 4-bit quantized, converted from
LetheanNetwork/lemmy's bf16
safetensors via mlx_lm.convert. Unmodified Google weights hosted
in the Lethean namespace so downstream tools don't have to depend
on external mlx-community mirrors.
For the LEK-merged sibling see lthn/lemmy.
License
Apache 2.0, subject to the Gemma Terms of Use.
- Downloads last month
- 7
Model size
25B params
Tensor type
BF16
·
U32 ·
Hardware compatibility
Log In to add your hardware
4-bit
Model tree for LetheanNetwork/lemmy-mlx
Base model
google/gemma-4-26B-A4B Finetuned
google/gemma-4-26B-A4B-it Quantized
LetheanNetwork/lemmy
Start the MLX server
# Install MLX LM: uv tool install mlx-lm# Start a local OpenAI-compatible server: mlx_lm.server --model "LetheanNetwork/lemmy-mlx"