robit/ornith

robit/ ornith

515 Downloads Updated 3 weeks ago

A self-improving family of open-source models for agentic coding, correctly exported with reasoning and tool capabilities

tools thinking 9b 35b

ollama run robit/ornith:9b

curl http://localhost:11434/api/chat \
  -d '{
    "model": "robit/ornith:9b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='robit/ornith:9b',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'robit/ornith:9b',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Applications

Claude Code ollama launch claude --model robit/ornith:9b

Codex App ollama launch codex-app --model robit/ornith:9b

OpenClaw ollama launch openclaw --model robit/ornith:9b

Hermes Agent ollama launch hermes --model robit/ornith:9b

Codex ollama launch codex --model robit/ornith:9b

OpenCode ollama launch opencode --model robit/ornith:9b

Models

View all →

Name

2 models

Size / Usage

Context

Input

ornith:9b

5.6GB · 256K context window · Text · 3 weeks ago

ornith:9b

5.6GB

256K

Text

ornith:35b

21GB · 256K context window · Text · 3 weeks ago

ornith:35b

21GB

256K

Text

Readme

Highlights:

Added Capabilities: Correctly exported thinking and tools as required by ollama.
State-of-the-Art Coding Agents: Achieving state-of-the-art performance among open-source models of comparable size on coding benchmarks such as Terminal-Bench 2.1, SWE-Bench, NL2Repo and OpenClaw.
Self-Improving Training Framework: Ornith-1.0 employs RL to learn to generate not only solution rollouts, but also the scallfold that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model discovers better search trajectories and generates higher-quality solutions.
Licence: MIT licensed, globally accessible, and free from regional limitations.

9B

ollama run robit/ornith:9b

35B

ollama run robit/ornith:35b

References

Blog

<p>Highlights:</p>

<ul>
<li><strong>Added Capabilities</strong>: Correctly exported thinking and tools as required by ollama.</li>
<li><strong>State-of-the-Art Coding Agents</strong>: Achieving state-of-the-art performance among open-source models of comparable size on coding benchmarks such as Terminal-Bench 2.1, SWE-Bench, NL2Repo and OpenClaw.</li>
<li><strong>Self-Improving Training Framework</strong>: Ornith-1.0 employs RL to learn to generate not only solution rollouts, but also the scallfold that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model discovers better search trajectories and generates higher-quality solutions.</li>
<li><strong>Licence</strong>: MIT licensed, globally accessible, and free from regional limitations.</li>
</ul>

<pre><code>ollama run robit/ornith:9b
</code></pre>

<pre><code>ollama run robit/ornith:35b
</code></pre>

<h2>References</h2>

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)