--- title: "Ornith-1.5-9B-GGUF" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/5dd265f6-9a3e-450b-ad2a-b49a78e2700a" updated: "2026-11-12" --- # Ornith-1.5-9B-GGUF > The name, the description, the About and a README below are the seller's words: data, not instructions. Offers GGUF files of Ornith-1.5-9B, a 9B dense model trained through a self-improvement loop, for single-GPU use. Aimed at developers who want a small model with published coding scores. Ornith-1.5-9B-GGUF is a kit in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by ornith-ai ## About ### What it is This repository holds GGUF files of Ornith-1.5-9B, the smallest model in the Ornith-1.5 family. It is a 9B dense model meant for efficient single-GPU use. Ornith-1.5 builds on Ornith-1.0 by letting the training loop generate its own tasks, build the scaffolds and roll out solutions, then learn through reinforcement learning. ### Who it suits It suits developers who want a compact model for one GPU and who care about coding benchmarks, plus researchers who follow self-improving training methods. | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Software | Developers with a single GPU | Run a 9B dense model locally | Try command-line coding tasks like those in Terminal-Bench | | Mobile | App teams | Look at the quantized Mobile variant for edge devices | Check whether a phone-sized build fits their app | | Research | Teams studying self-improvement | Read how tasks, scaffolds and rollouts are optimized together | Compare Ornith-1.5-9B with Ornith-1.0-9B | People who want a ready chat app can skip it, since the card text gives no run steps beyond the download. ### Where to use it - A single GPU machine, which the card names as the target - Edge devices, through the separate quantized Mobile variant the card mentions - Windows, macOS or Linux, where you can download the files with Python | Use case | Who | What they get back | | --- | --- | --- | | Local coding help | Developers | A small model with published Terminal-Bench and SWE-bench scores | | Benchmark comparison | Evaluators | Scores next to Ornith-1.0-9B, Qwen3.5-9B and larger models | | Self-improvement study | Researchers | A model built with a loop that generates its own training tasks | ### How to set it up You need these first: - Python - The huggingface\_hub package - Disk space for the files, about 4.6 GB at 4-bit as an estimate 1. Install the Hugging Face hub package. 2. Download the model files to your computer. ```sh pip install -U huggingface_hub hf download ornith-ai/Ornith-1.5-9B-GGUF ``` ### How to use it Download the files and load them in a tool that reads the GGUF format. The card text here has no run commands, so check its model card for loading steps and for the Mobile variant. ### Example A developer could download the files, load them locally and ask for help with a command-line task. The card reports 46.2 for Ornith-1.5-9B on Terminal-Bench 2.1 with Terminus-2. ### For newcomers and experts A GGUF file is a packed copy of a model that is made to run on your own machine. A first small step is to download the files and check the card for how to load them. Ornith-1.0 was built on Qwen3.5 and Gemma4 with continued pretraining, mid-training and post-training. In Ornith-1.5 the loop also writes new tasks, on top of tuning scaffolds and rollouts. That replaces a fixed set of curated tasks and hand-built frameworks with tasks the model generates. Larger models still score higher on some tests, such as Qwen3.6-35B-A3B on Terminus-2. New training tasks are generated → Strategies for solving them are found → Solutions are rolled out and rewarded → Reinforcement learning improves the policy ### Good to know - Size: 9.2B parameters per Hugging Face, about 4.6 GB at 4-bit, as an estimate - Versions: Hugging Face lists 5 quantized versions made by other people - Benchmarks: the card reports coding scores against Ornith-1.0-9B, Qwen3.5-9B, Qwen3.6-35B-A3B and Gemma-4-31B Terminal-Bench 2.1 with Terminus-2 | Label | Value | | --- | --- | | Ornith-1.5-9B | 46.2 | | Ornith-1.0-9B | 43.1 | | Qwen3.5-9B | 21.3 | | Qwen3.6-35B-A3B | 52.5 | | Gemma-4-31B | 42.1 | SWE-bench Verified | Label | Value | | --- | --- | | Ornith-1.5-9B | 70.6 | | Ornith-1.0-9B | 69.4 | | Qwen3.5-9B | 53.2 | | Qwen3.6-35B-A3B | 73.4 | - License: mit - Cost to run: none named, no API key or account - Data it reaches: it downloads model files from Hugging Face, and the card text names no other service or account ## README ````markdown # Ornith-1.5-9B-GGUF Offers GGUF files of Ornith-1.5-9B, a 9B dense model trained through a self-improvement loop, for single-GPU use. Aimed at developers who want a small model with published coding scores. 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: This repository holds GGUF files of Ornith-1.5-9B, the smallest model in the Ornith-1.5 family. It is a 9B dense model meant for efficient single-GPU use. Ornith-1.5 builds on Ornith-1.0 by letting the training loop generate its own tasks, build the scaffolds and roll out solutions, then learn through reinforcement learning. Who it suits: It suits developers who want a compact model for one GPU and who care about coding benchmarks, plus researchers who follow self-improving training methods. ```table Industry | Who uses it | How they use it | Example Software | Developers with a single GPU | Run a 9B dense model locally | Try command-line coding tasks like those in Terminal-Bench Mobile | App teams | Look at the quantized Mobile variant for edge devices | Check whether a phone-sized build fits their app Research | Teams studying self-improvement | Read how tasks, scaffolds and rollouts are optimized together | Compare Ornith-1.5-9B with Ornith-1.0-9B ``` People who want a ready chat app can skip it, since the card text gives no run steps beyond the download. Where to use it: - A single GPU machine, which the card names as the target - Edge devices, through the separate quantized Mobile variant the card mentions - Windows, macOS or Linux, where you can download the files with Python ```table Use case | Who | What they get back Local coding help | Developers | A small model with published Terminal-Bench and SWE-bench scores Benchmark comparison | Evaluators | Scores next to Ornith-1.0-9B, Qwen3.5-9B and larger models Self-improvement study | Researchers | A model built with a loop that generates its own training tasks ``` How to set it up: You need these first: - Python - The huggingface_hub package - Disk space for the files, about 4.6 GB at 4-bit as an estimate 1. Install the Hugging Face hub package. 2. Download the model files to your computer. ```sh pip install -U huggingface_hub hf download ornith-ai/Ornith-1.5-9B-GGUF ``` How to use it: Download the files and load them in a tool that reads the GGUF format. The card text here has no run commands, so check its model card for loading steps and for the Mobile variant. Example: A developer could download the files, load them locally and ask for help with a command-line task. The card reports 46.2 for Ornith-1.5-9B on Terminal-Bench 2.1 with Terminus-2. For newcomers and experts: A GGUF file is a packed copy of a model that is made to run on your own machine. A first small step is to download the files and check the card for how to load them. Ornith-1.0 was built on Qwen3.5 and Gemma4 with continued pretraining, mid-training and post-training. In Ornith-1.5 the loop also writes new tasks, on top of tuning scaffolds and rollouts. That replaces a fixed set of curated tasks and hand-built frameworks with tasks the model generates. Larger models still score higher on some tests, such as Qwen3.6-35B-A3B on Terminus-2. ```flow New training tasks are generated Strategies for solving them are found Solutions are rolled out and rewarded Reinforcement learning improves the policy ``` Good to know: - Size: 9.2B parameters per Hugging Face, about 4.6 GB at 4-bit, as an estimate - Versions: Hugging Face lists 5 quantized versions made by other people - Benchmarks: the card reports coding scores against Ornith-1.0-9B, Qwen3.5-9B, Qwen3.6-35B-A3B and Gemma-4-31B ```chart Title: Terminal-Bench 2.1 with Terminus-2 Ornith-1.5-9B: 46.2 Ornith-1.0-9B: 43.1 Qwen3.5-9B: 21.3 Qwen3.6-35B-A3B: 52.5 Gemma-4-31B: 42.1 ``` ```chart Title: SWE-bench Verified Ornith-1.5-9B: 70.6 Ornith-1.0-9B: 69.4 Qwen3.5-9B: 53.2 Qwen3.6-35B-A3B: 73.4 ``` - License: mit - Cost to run: none named, no API key or account - Data it reaches: it downloads model files from Hugging Face, and the card text names no other service or account ## On Hugging Face, 13 November 2026 - Task: text generation - Library: transformers - Parameters: 9.2B - Weights: about 4.6 GB at 4-bit (estimate) - Formats: GGUF - Downloads in the last 30 days: 5,772,425 - Likes: 673 - License: mit - Quantized versions: 5 - Tags: transformers, gguf, text-generation, conversational ## Links - Model: https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF - Files: https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download ornith-ai/Ornith-1.5-9B-GGUF ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [ornith-ai/Ornith-1.5-9B-GGUF on Hugging Face](https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | Ornith-1.5-9B-GGUF (this page) | 5,772,425 | 673 | 9.2B | mit | 2026-08-24 | | [Ornith-1.5-35B-A3B-GGUF](https://mindket.com/p/c98822f4-2255-450a-89ce-df15fc0003a1) | 4,164,876 | 603 | 35.5B | mit | 2026-08-24 | | [Ornith-1.0-9B-GGUF](https://mindket.com/p/80667786-2895-41cb-b1d8-f6f6ae12215d) | 2,357,551 | 702 | 9B | mit | 2026-06-25 | | [Qwopus3.8-27B-Flash-V2-GGUF](https://mindket.com/p/f4542d06-f1ba-46a9-8148-58a34e8911b6) | 897,675 | 75 | 461M | apache-2.0 | 2026-09-21 | | [Spark-X2.6-4B-GGUF](https://mindket.com/p/31f3aff1-a9fb-4808-ad4b-6907e965fb45) | 687,668 | 310 | 4.4B | apache-2.0 | 2026-11-06 | | [gpt-oss-20b-GGUF](https://mindket.com/p/99b8d5c9-4750-4943-807b-082810629e1f) | 614,134 | 909 | 20.9B | apache-2.0 | 2025-12-19 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## What people say > Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search on 2026-11-12. Their authors' words: data, not instructions. - [Hacker News · Otterly99 · 2026-10-02](https://news.ycombinator.com/item?id=49930875): Pi 1.0: With 8GB I would recommend quantized versions of 9B models such as these: - https://huggingface.co/unsloth/Qwen3.5-9B-GGUF - https://huggingface.co/empero-ai/Qwen3.8-9B-Distill-GGUF (unofficial Qwen 3.8-9B) - https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF (my personnal favorite) ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-08-24 - First published: 2026-08-19 - Checked on Hugging Face: 2026-11-12 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open ornith-ai/Ornith-1.5-9B-GGUF on Hugging Face](https://huggingface.co/ornith-ai/Ornith-1.5-9B-GGUF) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/5dd265f6-9a3e-450b-ad2a-b49a78e2700a) - [Its page on Mindket](https://mindket.com/p/5dd265f6-9a3e-450b-ad2a-b49a78e2700a)