Ornith-1.5: From Self-Scaffolding to Self-Improvement

Ornith-1.5 is a new foundation model series that advances self-improvement by enabling models to autonomously generate tasks, create solution strategies, and learn through reinforcement learning without relying on fixed human-curated datasets. Available in three sizes (397B MoE, 35B MoE, and 9B dense), Ornith-1.5 achieves state-of-the-art performance among open-source models, with the flagship 397B version matching Claude Opus 4.8 on key benchmarks while the compact 9B version can run on mobile devices and outperforms much larger competitors.