Promote Sign in Sign up

Multimodal model

Ornstein Hermes 3.6 27b MLX 8bit

Ornstein-Hermes-3.6-27b-MLX-8bit is Gestalt Labs’ 8-bit MLX quantization of Ornstein-Hermes-3.6-27b, a Hermes-format function-calling fine-tune of Qwen 3.6 27B multimodal. It is optimized for Apple Silicon, supports image-text-to-text use, and targets agentic tool use with near-lossless 8-bit compression.

GestaltLabs Multimodal Open weights apache-2.0
Open model card ↗
Type
Multimodal
Weights
Open
Downloads
131
Likes
4
Pipeline
image text to text
License
apache-2.0

Model hub

GestaltLabs/Ornstein-Hermes-3.6-27B-MLX-8bit

Repo
GestaltLabs/Ornstein-Hermes-3.6-27B-MLX-8bit
Library
mlx
Access
Public
Technical tags and languages
en mlx safetensors qwen3_5 apple-silicon quantized 8-bit qwen3.6 multimodal function-calling tool-use agentic hermes
Open on Hugging Face

More from GestaltLabs

3

About GestaltLabs

Independent Canadian open-source AI research lab developing and releasing fully permissive open-weight reasoning models, multimodal assistants, and local deployment tools. Core model lines include Ornstein (multimodal/MoE reasoning), CHIMERA (self-correction), and SABER (refusal ablation), alongside bio-inspired vision architectures (MPKnet).