# Piccolo: Co-Evolving Decision Models and Data through Recursive Self-Improvement

> Source: <https://aiflash.com/news/154189/>
> Published: 2026-11-19 08:01:08+00:00

**Key takeaway:** Piccolo spans 0.8B to 27B parameters and predicts probabilities over candidate answers for text and visual judgments, pairing model training with Agentic Data Engineering and recursive self-improvement to co-evolve data and capabilities.

We present Piccolo, a family of multimodal decision models (0.8B to 27B) that coevolves model capabilities and training data via Agentic Data Engineering and recursive self-improvement (RSI). The models predict probabilities over candidate answers for general-purpose text and visual judgments, agent
