OpenComputer: Verifiable Software Worlds for Computer-Use Agents
Summary
OpenComputer presents a framework for creating verifiable software environments for computer-use agents, integrating state verifiers, self-improving verification layers, task synthesis, and evaluation systems across 33 desktop applications. Experiments show its verifiers align better with human judgment than LLM-as-judge, and frontier agents struggle with end-to-end completion.
View Cached Full Text
Cached at: 05/20/26, 02:35 AM
Paper page - OpenComputer: Verifiable Software Worlds for Computer-Use Agents
Source: https://huggingface.co/papers/2605.19769
Abstract
OpenComputer presents a framework for creating verifiable software environments for computer-use agents through integrated state verification, self-improving layers, task synthesis, and evaluation systems across multiple desktop applications.
We present OpenComputer, averifier-grounded frameworkfor constructing verifiable software worlds for computer-use agents. OpenComputer integrates four components: (1) app-specificstate verifiersthat expose structured inspection endpoints over real applications, (2) aself-evolving verification layerthat improves verifier reliability using execution-grounded feedback, (3) atask-generation pipelinethat synthesizes realistic and machine-checkable desktop tasks, and (4) anevaluation harnessthat records full trajectories and computes auditablepartial-credit rewards. In its current form, OpenComputer covers 33desktop applicationsand 1,000 finalized tasks spanning browsers, office tools, creative software, development environments, file managers, and communication applications. Experiments show that OpenComputer’s hard-coded verifiers align more closely withhuman adjudicationthanLLM-as-judge evaluation, especially when success depends on fine-grained application state. Frontier agents struggle with end-to-end completion despite partial progress, and open-source models exhibit sharp drops from their OSWorld-Verified scores, exposing a persistent gap in robustcomputer automation.
View arXiv pageView PDFGitHubAdd to collection
Get this paper in your agent:
hf papers read 2605\.19769
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2605.19769 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.19769 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.19769 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
OpenComputer | An Open Source Computer Built For Agents.
OpenComputer is an open-source virtual machine environment for AI agents that provides a human-accessible computer interface, allowing agents to safely operate while users can observe and collaborate. It runs locally with small context models and avoids screenshot-based navigation for efficiency.
Open Computer Use
Open Computer Use is an open-source MCP (Model Context Protocol) for AI agents to control computer interfaces.
Training Needs Trustworthy Worlds: Verified Synthetic Web Environments for Agent Learning
This paper introduces a framework for constructing verified synthetic web environments to improve the training of web agents, demonstrating enhanced performance and transferability across benchmarks.
OSWorld2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks
OSWorld 2.0 is a new benchmark for evaluating computer-use agents on 108 long-horizon, real-world workflows. Current agents like Claude Opus 4.8 and GPT-5.5 achieve low completion rates, highlighting significant limitations in handling complex, multi-step tasks.
OpenLife: Toward Open-World Artificial Life with Autonomous LLM Agents
This paper introduces OpenLife, a proof-of-concept system that uses autonomous LLM agents with persistent memory and budget-based metabolism to realize open-world artificial life. Experiments over twelve weeks show emergent life-like dynamics including spontaneous activity, individuation, and social structure.