# OpenAI Launches BrowseComp Benchmark for Browsing Agents

OpenAI has announced BrowseComp, a new benchmark designed to evaluate browsing agents—AI systems that can navigate and interact with websites autonomously.

The benchmark provides a standardized way to measure how well AI agents perform web-based tasks, such as finding information, filling out forms, or completing multi-step processes across different websites. This addresses a critical gap in AI evaluation, as browsing capabilities become increasingly important for practical AI applications.

BrowseComp matters because it establishes clear metrics for an emerging AI capability. As companies develop agents that can browse the web on behalf of users—booking flights, researching products, or gathering information—having reliable benchmarks ensures these systems can be properly tested and compared.

The release signals growing industry focus on agentic AI systems that can perform complex, real-world tasks rather than simply responding to prompts