Tag
pi-fusion is an extension of pi that improves performance at lower cost by parallel fan-out of multiple models and fusing results, supporting prompt rewriting and session archiving.
DiffusionGemma, a 26B MoE model based on Gemma 4, achieves over 1000 tokens per second using diffusion for text generation in 256-token blocks, fitting in 18GB VRAM with quantization, released under Apache 2.0.
Visual Para-Thinker++ proposes a single-policy multi-agent framework for visual reasoning that uses role-conditioned agents (Main, Worker, Summary) and dedicated training methods to reduce hallucinations and improve efficiency, outperforming baselines on hallucination-sensitive benchmarks.