UiPath Claims AI Agent Crown with Claude Opus 4.5 on Key Benchmark

NEW YORK, NY – January 14, 2026 – Automation leader UiPath has secured the top position in a highly competitive field of artificial intelligence, with its Screen Agent, powered by Anthropic's Claude Opus 4.5, achieving the number one rank on the OSWorld-Verified benchmark. This achievement signals a significant milestone in the race to develop AI agents capable of autonomously performing complex computer tasks, providing enterprises with the validation needed to deploy AI at scale.

The OSWorld benchmark, an independent evaluation conducted by the OSWorld research group, is widely regarded as one of the most rigorous tests for agentic AI. The ranking validates UiPath's technology against a backdrop of intense competition from general-purpose models and specialized agentic frameworks, reinforcing its position in the rapidly evolving enterprise automation market.

The New Gold Standard for AI Agents

For years, the true capability of AI agents has been difficult to measure, often tested in simulated or confined environments that fail to capture the complexity of real-world business operations. The OSWorld-Verified benchmark was created to address this gap, providing a unified, interactive computer environment to assess an AI agent's ability to handle open-ended tasks that span arbitrary applications and operating systems.

Unlike previous evaluations, OSWorld uses real virtual machines (VMs) running on Windows, Ubuntu, and macOS. Agents are tested on their ability to successfully complete 369 distinct computer tasks derived from real-world use cases. These tasks are not simple, single-step actions; they often involve intricate workflows that require navigating web and desktop applications, performing OS file operations, and integrating information across multiple programs. The benchmark's design is a direct reflection of the messy, unpredictable digital environments that knowledge workers navigate daily.

The difficulty of this benchmark is underscored by historical performance data. While human experts can accomplish approximately 72.4% of the tasks, the best AI agents previously struggled to surpass a 12.2% success rate, often failing due to challenges in understanding graphical user interfaces (GUI) and lacking the operational knowledge to proceed. UiPath’s ascent to the top rank signifies a substantial leap in AI capability, particularly in the agent's ability to perceive, reason, and act reliably within standard enterprise IT environments.

“Organizations need the confidence that their large-scale commitments to AI will pay off, which is where benchmarks can be incredibly helpful in validating specific use cases and critical workflows,” said Mircea Neagovici-Negoescu, Senior Vice President of AI and Research at UiPath, in a statement. This sentiment is echoed across an industry grappling with how to move AI from experimental pilots to production-ready systems.

A New Leader in the AI Agent Arms Race

UiPath's top ranking is not an overnight success but the result of sustained investment and strategic partnerships. The achievement builds on a previous milestone in September 2025, when the UiPath Screen Agent, then powered by OpenAI's GPT-5, secured the number two position on the same benchmark. The switch to and subsequent success with Claude Opus 4.5 highlights UiPath's vendor-agnostic strategy, focusing on integrating the best-performing large language models (LLMs) into its automation platform.

Claude Opus 4.5, Anthropic's flagship model, brings several key advancements that are particularly well-suited for agentic automation. It features enhanced computer-use capabilities, including a novel "zoom" action that allows the agent to inspect small UI elements and fine print with high resolution. Furthermore, its ability to manage long-context conversations and maintain reasoning across multi-step workflows makes it a powerful engine for orchestrating complex tasks. This allows the AI agent to not just execute a command, but to plan, troubleshoot, and adapt its approach in real-time.

The competitive landscape for agentic AI is heating up, with tech giants like Google, Microsoft, and OpenAI all investing heavily in creating autonomous systems. However, UiPath's strategy differentiates itself by focusing on orchestration and enterprise-grade governance. Through platforms like UiPath Maestro, the company aims to create a cohesive system where AI agents, traditional software robots, and human employees can collaborate within a single, managed workflow. This approach allows businesses to leverage the power of cutting-edge AI models while maintaining the security, control, and compliance required for sensitive business processes.

From Robotic Process to Autonomous Action

The rise of agentic AI represents a fundamental paradigm shift, moving beyond the rules-based limitations of traditional Robotic Process Automation (RPA). While RPA excels at automating repetitive, predictable tasks, agentic AI introduces the ability to handle probabilistic, unstructured work. These AI agents can understand user intent from natural language, devise a plan to achieve a goal, and execute that plan across multiple systems, learning and optimizing as they go.

This technology is the core of UiPath ScreenPlay, a platform designed to scale automation with adaptive intelligence. For businesses, the potential is transformative. “Having had an early look at UiPath ScreenPlay, we’re excited about its potential to meaningfully improve how we scale automation,” noted Noble Keyser, manager of Enterprise AI and Automation at SimpleTire. “Its adaptive intelligence could support our growing partner ecosystem while helping reduce ongoing maintenance so our teams can stay focused on growth.”

Enterprise adoption is already well underway. According to Gartner, one-third of all generative AI interactions will involve autonomous agents by 2028. Use cases are emerging across every department, from finance teams automating complex reconciliations to IT departments deploying agents for autonomous incident response. By handling the complex, multi-system digital work that consumes countless hours, these agents promise to unlock significant productivity gains and free human workers to focus on strategic, creative, and high-value initiatives.

Navigating the Hurdles of Enterprise Adoption

Despite the immense promise, the path to enterprise-wide agentic AI is fraught with challenges. Chief among them are concerns over governance, security, and reliability. Handing over the keys to complex business processes to an autonomous AI agent requires a profound level of trust, and many leaders are wary of the "black box" problem, where an AI's decision-making process is opaque.

Furthermore, integration with legacy systems, ensuring data privacy, and demonstrating a clear return on investment remain significant barriers. The market is also experiencing a wave of "agent washing," where simple chatbots or assistants are rebranded as agentic AI, creating confusion and skepticism.

This is precisely why independent, rigorous benchmarks like OSWorld are so critical. They cut through the marketing hype and provide tangible proof of an agent's capabilities. UiPath's No. 1 ranking serves as a powerful proof point for enterprises that are, as Neagovici-Negoescu stated, daunted by the prospect of investing in AI “at enterprise speed and scale.”

By building its platform around principles of controlled agency, developer flexibility, and seamless integration, UiPath aims to provide the guardrails necessary for safe and confident scaling. Its focus on security, governance, and interoperability is designed to assure businesses that they can harness the power of autonomous agents without sacrificing control or exposing themselves to unacceptable risk. This validated performance on a challenging, real-world benchmark provides a crucial signal that the technology is maturing, offering a reliable path for businesses to transition into a future where automation can finally deliver on the full potential of artificial intelligence.