# ornith-ai/Ornith-1

Repository: https://github.com/ornith-ai/Ornith-1
Canonical: https://ross.abutalabs.com/products/ornith-1
License: MIT
License Family: permissive
Last push: 2026-08-22T20:03:57+00:00

## Health v2 (maintenance only)
Score: 58/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 35, longevity 5
- inputs: {"age_days": 73, "days_push": 11, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1958, forks 191 (observed 2026-08-28T04:05:58.894003+00:00)

## What it is
Ornith is a family of open-weight large language models (397B scale) designed for agentic tasks, released under MIT license with weights hosted on HuggingFace. The repo documents two generations: Ornith-1.0, which jointly optimizes task scaffolds and solution rollouts, and Ornith-1.5, which adds a full self-improvement loop that proposes tasks, generates scaffolds, and produces rollouts for reinforcement learning.

## Use cases
- run an open-weight model for autonomous agent tasks
- fine-tune or extend a self-improving agentic LLM
- compare open models against proprietary frontier models on agent benchmarks
- study self-scaffolding and RL-based self-improvement for LLMs
- deploy a permissively licensed large model without regional restrictions

## When to choose
- you need open, MIT-licensed model weights for agentic workloads
- you want to research self-improvement loops and scaffold optimization
- you need a large open model as an alternative to closed frontier models

## When to avoid
- you need a small model that runs on consumer hardware
- you want a ready-made agent application rather than model weights
- you lack the GPU infrastructure to serve a 397B-parameter model

## Facets
- artifact type: dataset
- maturity: active
- function: llm-training, agent-framework, machine-learning, llm-inference
- domain: large-language-models, reinforcement-learning, machine-learning
- platform: python, cross-platform
- tags: model-weights, self-improvement, agentic-tasks, scaffolding, open-weights, mit-license, huggingface, ai-agents, gpu

## Member repositories
- ornith-ai/Ornith-1 (main) score 58

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:58.894003+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:06:00.043309+00:00, confidence not recorded.
  - readme: https://github.com/ornith-ai/Ornith-1 (fetched 2026-08-28T04:05:58.894003+00:00, sha a4daa880926a)
- Data as of 2026-08-30T08:39:29.467469+00:00.