Orchard: An open framework for scalable agentic AI

Researchers have announced the launch of Orchard, an open-source framework designed to enhance agentic AI research. Orchard is built around a reusable environment service that supports training and evaluating agents across various domains. For instance, the Orchard-SWE model achieved a score of 69.7% on the SWE-bench test, demonstrating the effectiveness of smaller models on complex tasks.