Ornith-1.5: From Self-Scaffolding to Self-Improvement
Summary
Ornith-1.5 introduces a major upgrade to its self-improvement framework, extending self-scaffolding into a full self-improvement loop where the model proposes tasks, generates task-specific scaffolds, and produces solution rollouts for reinforcement learning. The release covers three model scales (397B MoE, 35B MoE, 9B Dense) and reports state-of-the-art open-source performance across reasoning, coding, and agentic benchmarks, including edge deployments for the 9B variant.