Pistis AI Can Search the Web and Take Actions For You

A new technical report introduces the Pistis model family: 27B- and 9B-parameter multimodal large language models built on Qwen3.6 and Qwen3.5, respectively, and developed through a general and scalable post-training framework. That framework first establishes a foundation with large-scale multimodal supervised fine-tuning, then adds Interleaved Distillation and Reinforcement Learning (IDRL), a post-training paradigm that alternates on-policy distillation and reinforcement learning within a single training loop. At both model scales, the framework produces Pistis-Thinking, designed to strengthen deep multimodal reasoning, and Pistis-Agentic, which incorporates agentic trajectory data to support long-horizon planning, iterative reasoning, and tool use. Pistis-Agentic is particularly strong in multimodal search, and both scales outperform their corresponding base models. The report also presents Pistis-Auto-Harnessing (PAH), a system-level method that automatically improves the agent's inference harness through iterative optimization, enhancing model performance without updating the model parameters or increasing the interaction budget.