NOEMA
Private · local · intelligence
NoemaAI Labs on Hugging Face

Private intelligence,
entirely yours.

Open-weight models and local-first AI for Apple devices, built for private, resource-conscious inference.

Explore Noema  ↗ Browse models
Featured work

Built to stay close.

01 · Latest model

Noema 1.5 2B

A balanced local 2B model for instructions, multi-turn work, code, and efficient reasoning.

Transformers · text generation  ↗
02 · Experimental runtime

Noema Overfit

Paged MoE packages for models whose complete weights exceed device memory.

GGUF · expert paging  ↗
03 · Grounded retrieval

Knowledge Packs

Public-domain reference text prepared for Noema's on-device retrieval.

Dataset · local retrieval  ↗

Noema 1.5, your format.

Native GGUF MLX BF16 MLX 8-bit MLX 6-bit MLX 4-bit

Open-weight unless a repository says otherwise. Repository terms are authoritative.

Documentation ↗ GitHub ↗ Get Noema ↗