Ornith-1.5 is an open-source foundation model family (397B MoE, 35B MoE, and 9B dense) trained through a self-improvement loop in which the model generates its own tasks, scaffolds, and solution rollouts for reinforcement learning, targeting reasoning, coding, and agentic workloads.
It is most valuable when developers need a high-performing open-source coding or agentic model without proprietary API dependencies, or when building mobile/edge applications requiring local AI inference with strong coding and reasoning capabilities.