Ornith

2 aliases · Ornith-1.5 9B (dense) + 35B-A3B (MoE) · MIT license · served day-0, reasoning and tool calling working out of the box.

Ornith-1.5 is an open, MIT-licensed family from ornith-ai, added to rapid-mlx the day it appeared. Both models are built on the Qwen3.5-class GatedDeltaNet architecture, so they slot into the engine's existing hybrid-cache serving path: the hermes tool-call envelope and the qwen3 reasoning parser work with no extra config, and thinking streams as reasoning_content.

Both aliases are bf16 — full precision. The 9B is a dense checkpoint; the 35B-A3B is a mixture-of-experts that activates about 3B parameters per token, which keeps decode brisk for its weight class. Both are mirrored on our CDN for fast pulls.

family

Ornith (ornith-ai)

aliases

2

license

MIT

parsers

hermes · qwen3

OpenAI base URL

http://localhost:8000/v1

Download

Every alias on this page downloads with one command — the pull buttons in the tables below copy it. All 2 aliases on this page are mirrored on the rapid-mlx CDN — our fresh-install test measured CDN pulls at 90 MB/s, with automatic mid-pull fallback to Hugging Face if a mirror file slows down. Weights land in the standard Hugging Face cache, and rapid-mlx serve pulls automatically on first use. Live mirror status →

Usage

rapid-mlx serve ornith-1.5-9b-bf16

curl http://localhost:8000/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{"model":"ornith-1.5-9b-bf16","messages":[{"role":"user","content":"Hello"}]}'

Aliases

Frequently asked questions

What is Ornith-1.5?

An open, MIT-licensed family from ornith-ai, served day-0 in rapid-mlx 0.13.0 — a 9B dense model and a 35B-A3B MoE on the Qwen3.5-class GatedDeltaNet architecture, with tool calling and reasoning parsed out of the box.

How much RAM do the Ornith models need?

Both ship bf16. The 9B (16.7 GB of weights) wants a 32 GB Mac; the 35B-A3B (69.3 GB) wants 96 GB or more.

Where next