Ornith-1.0: Self-scaffolding LLMs for agentic coding

Ornith-1.0 is a new family of open-source models ranging from 9B to 397B parameters designed for agentic coding tasks, achieving state-of-the-art performance on coding benchmarks. Its key innovation is a self-improving training framework where the model jointly learns to generate solutions and the task-specific scaffolds that guide those solutions, rather than relying on fixed human-designed harnesses. The 397B variant outperforms Claude Opus 4.7 on major benchmarks, while the compact 9B version matches much larger models, demonstrating strong coding capabilities across different deployment scales.