Ornith-1.5-35B-A3B-GGUF

Ornith-1.5-35B-A3B-GGUF

Runs a 35B mixture-of-experts text model tuned for coding and agent work, using only about 3B parameters per token. This repository holds its GGUF files for local use.

Made by ornith-ai on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: Ornith-1.5-35B-A3B is a text generation model from ornith-ai, and this repository holds it in GGUF format. It is a mixture-of-experts model, so only about 3B of its 35.5B parameters work on each token.

It extends Ornith-1.0, which was built on Qwen3.5 and Gemma4. Version 1.5 adds a training loop where the model generates its own tasks, builds scaffolds and learns from solution attempts through reinforcement learning.

Who it suits: Developers who want a coding and agent model they can keep on their own machine, and people who test local models against coding benchmarks.

Industry | Who uses it | How they use it | Example
Software | Developers and tool builders | Run it locally as a coding assistant | Ask for a fix to a failing script
DevOps | Teams that automate terminal work | Try it as the model behind a terminal agent | Plan the steps of a command line task
Research | People comparing open models | Score it on coding and agent tests | Compare it with other models of similar size
Hobby | Local model enthusiasts | Load the GGUF file in a local runner | Chat about code without a hosted service

Skip it if your computer has little free disk space, since the 4-bit weights come to about 17.8 GB by Hugging Face estimate.

Where to use it:

• Windows, macOS or Linux with Python, to download the files
• A local runner that reads GGUF files
• Terminal agent setups, which the card used for its coding benchmarks

Use case | Who | What they get back
Coding help | Developers | Code and explanations from a model that activates few parameters per token
Agentic coding | Agent builders | A model the authors scored on terminal and agent tasks
Model comparison | Researchers | A GGUF copy to test against other open models

How to set it up:

• Python and pip to run the download commands
• About 17.8 GB of free disk space for the 4-bit weights
• A local runner that can open GGUF files

1. Install the Hugging Face tools.
2. Download the files from this repository.
3. Open the GGUF file in your local runner. The model card page has the run notes.

pip install -U huggingface_hub
hf download ornith-ai/Ornith-1.5-35B-A3B-GGUF

How to use it: Load the downloaded file in your runner, then chat with it or point a coding agent at it. The source shows no run command, so follow your runner for the exact steps.

Example: You ask it for a fix to a failing command line script and paste the error. You can expect a text reply with an explanation and code, since the authors tuned it for coding and agent tasks.

For newcomers and experts: If you are new to local models, think of the GGUF file as the model packed to run on your own computer. Download it, load it in a runner, and start with a short coding question.

If you know the field, this is a sparse mixture-of-experts model with 35.5B parameters and about 3B active per token. The authors say Ornith-1.5 jointly optimizes task generation, scaffold construction and solution rollouts, instead of using fixed human-made tasks and hand-built setups.

Download the GGUF file
Load it in a local runner
Send a coding or agent prompt
Read the generated reply

Good to know:

• The authors report beating Qwen 3.6-35B on their coding and agentic benchmarks, and these numbers come from them
• Other people have published 3 quantized versions and no fine-tuned versions, according to Hugging Face

Title: Terminal-Bench 2.1 score with the Terminus-2 setup
Ornith-1.5-35B-A3B: 67.8 points
Ornith-1.0-35B-A3B: 64.2 points
Qwen3.5-397B: 53.5 points
Qwen3.6-35B-A3B: 52.5 points
Muse-Glimmer-30B: 51.7 points
Gemma-4-31B: 42.1 points

• License: MIT
• Cost to run: none named
• Data it reaches: the files you download and the prompts you give it, with no other access named
• Not for: the source states no restrictions

On Hugging Face, 13 November 2026

• Task: text generation
• Library: transformers
• Parameters: 35.5B
• Weights: about 17.8 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 4,164,876
• Likes: 603
• License: mit
• Quantized versions: 3
• Tags: transformers, gguf, text-generation, conversational

Links

• Model: https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF
• Files: https://huggingface.co/ornith-ai/Ornith-1.5-35B-A3B-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download ornith-ai/Ornith-1.5-35B-A3B-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.