Orchard: An Open-Source Agentic Modeling Framework
:::info Stub — Full Engineering Breakdown Coming This paper has a linked code implementation and was featured on Hugging Face Papers with 14 upvotes. A full breakdown with production viability rating, implementation notes, and honest limitations is being written. Subscribe to AI Letters → :::
| Authors | Baolin Peng et al. |
| Year | 2026 |
| HF Upvotes | 14 |
| arXiv | 2605.15040 |
| Download | |
| Code | https://github.com/microsoft/Orchard |
Abstract
Agentic modeling aims to transform LLMs into autonomous agents capable of solving complex tasks through planning, reasoning, tool use, and multi-turn interaction with environments. Despite major investment, open research remains constrained by infrastructure and training gaps. Many high-performing systems rely on proprietary codebases, models, or services, while most open-source frameworks focus on orchestration and evaluation rather than scalable agent training. We present Orchard, an open-source framework for scalable agentic modeling. At its core is Orchard Env, a lightweight environment service providing reusable primitives for sandbox lifecycle management across task domains, agent harnesses, and pipeline stages. On top of Orchard Env, we build three agentic modeling recipes. Orchard-SWE targets coding agents. We distill 107K trajectories from MiniMax-M2.5 and Qwen3.5-397B, introduce credit-assignment SFT to learn from productive segments of unresolved trajectories, and apply Balanced Adaptive Rollout for RL. Starting from Qwen3-30B-A3B-Thinking, Orchard-SWE achieves 64.3% on SWE-bench Verified after SFT and 67.5% after SFT+RL, setting a new state of the art among open-source models of comparable size. Orchard-GUI trains a 4B vision-language computer-use agent using only 0.4K distilled trajectories and 2.2K open-ended tasks. It achieves 74.1%, 67.0%, and 64.0% success rates on WebVoyager, Online-Mind2Web, and DeepShop, respectively, making it the strongest open-source model while remaining competitive with proprietary systems. Orchard-Claw targets personal assistant agents. Trained with only 0.2K synthetic tasks, it achieves 59.6% pass@3 on Claw-Eval and 73.9% when paired with a stronger ZeroClaw harness. Collectively, these results show that a lightweight, open, harness-agnostic environment layer enables reusable agentic data, training recipes, and evaluations across domains.
Engineering Breakdown
Plain English
Orchard is an open-source framework that addresses the gap between research and production for building autonomous AI agents. The paper identifies that while proprietary systems (like OpenAI's o1 or Claude with tool use) dominate, there's no scalable open-source infrastructure for training agents end-to-end—most open frameworks only handle orchestration and evaluation, not the actual training pipeline.
Key Engineering Insight
The core contribution is Orchard Env, a lightweight sandbox lifecycle manager that provides reusable primitives across different task domains and pipeline stages, letting teams train agents at scale without rebuilding infrastructure for each new task or environment.
Why It Matters for Engineers
Most teams building production agents today either use closed APIs or bolt together fragile orchestration scripts. Orchard removes the infrastructure tax by standardizing how environments are managed, sandboxed, and integrated into training loops—this directly reduces time-to-deployment for agents that need tool use, planning, and multi-turn reasoning.
Research Context
The agentic AI space exploded after ChatGPT and function calling, but open-source lagged behind proprietary systems because training agents requires both scalable infrastructure and task diversity. Orchard fills this gap by providing the foundational plumbing (Orchard Env) plus reference recipes for agent training, enabling the community to close the research-to-production gap that currently favors companies with internal infrastructure.
:::tip Subscribe Get weekly breakdowns of papers like this in AI Letters - the newsletter for engineers building production AI systems. :::