← Library · Frontier

DeepReinforce Releases Ornith-1.0, an AI Model that Writes Its Own Agent Scaffolds

DeepReinforce has launched Ornith-1.0, an open-source family of coding models (9B, 31B, 35B MoE, 397B MoE) built on Gemma 4 and Qwen 3.5, and released under the MIT license. Ornith-1.0 learns to generate its own orchestration logic, or 'scaffold', while simultaneously producing solutions. This novel self-scaffolding approach allows the model to co-evolve the scaffold with its policy during reinforcement learning.

Why it matters

This approach introduces 'self-scaffolding' as a new scaling axis in AI, enabling more adaptive and less black-box agentic coding, which could significantly advance the development of autonomous coding agents.

Learn one new AI thing every day.

Daily Deck sends you seven plain-English cards like this every morning. Free.

Start free