The A.I. Daily
Ornith-1.5-122B released: self-improving model built on Qwen
Ornith AI released Ornith-1.5-122B, which extends self-improvement from Ornith-1.0 (built on Qwen3.5 and Gemma4) to jointly optimize task generation, scaffold construction, and solution rollouts via reinforcement learning. The model is available in GGUF format on Hugging Face.