What is Ornith-1.0?
Open-weight coding models from 9B to 397B that learn to write their own agent scaffolds.
Who is Ornith-1.0 best for?
Ornith-1.0 is best for teams that want to self-host an open coding model for agent work, in sizes from 9B to 397B.
Self-hosted coding agents
The models ship with vLLM and SGLang recipes and an OpenAI-compatible server that returns tool calls.
Small hardware budgets
The 9B dense model is built for edge deployment, and the maker reports 69.4 on SWE-Bench Verified for it.
Teams that need a permissive license
MIT licensing with no regional limits lets companies adapt and ship the models commercially.
Researchers studying self-improvement
The launch post explains how the model learns its own task scaffolds and how reward hacking is held back.
How much does Ornith-1.0 cost?
Ornith-1.0 is free: the weights for every size are on Hugging Face under the MIT license, and the maker offers no paid API, so the only cost is the hardware you run it on.
- 9B, 35B and 397B weights on Hugging Face
- FP8 and GGUF versions too
- No paid API from the maker
Running the 397B model takes serious hardware: the maker's serving recipes assume one node with eight 80GB GPUs. A newer Ornith-1.5 family is also out from the same team. Full pricing on huggingface.co. We checked these prices on .
How do readers rate Ornith-1.0?
No one has rated Ornith-1.0 yet. Be the first.
Sign in to rate. One rating per account.
Is Ornith-1.0 open source?
Yes. Ornith-1.0's code is public under the MIT licence, so you can run it on your own server.
What are the best Ornith-1.0 alternatives?
Ornith-1.0 is an open source alternative to Claude Opus 4.7 and Muse Spark 1.3. DeepSeek V4.1 Flash and GLM-5.3-Flash do the same job and are open source too.
Anthropic's Opus 4.7 with a 1M context, still active for coding, agents and vision.
Meta's model for coding and long-running agents, on the API.
DeepSeek's small, efficient open model.
Z.ai's multimodal GLM-5.3-Flash with a 1M-token context.
Questions about Ornith-1.0
What is Ornith-1.0 built on?
Ornith-1.0 is post-trained on top of pretrained Gemma 4 and Qwen 3.5 models. Its makers trained it with reinforcement learning in which the model writes both the solution and the scaffold that guides it.
Who is Ornith-1.0 for?
Ornith-1.0 is best for teams that want to self-host an open coding model for agent work, in sizes from 9B to 397B.
Is Ornith-1.0 free?
Yes. Ornith-1.0 has a free Open weights plan: 9B, 35B and 397B weights on Hugging Face.
Is Ornith-1.0 open source?
Yes. Ornith-1.0's code is public under the MIT licence, so you can run it on your own server.
What are the best Ornith-1.0 alternatives?
The closest alternatives on Recaply are Claude Opus 4.7, Muse Spark 1.3, DeepSeek V4.1 Flash and GLM-5.3-Flash.
Where can I try Ornith-1.0?
On its official website, ornith.ai.







