Hugging Face has released Open Computer Agent, a free, cloud-hosted AI tool that can perform tasks on a virtual Linux machine preloaded with applications like Firefox. The agent, accessible via the web, can be prompted to complete tasks such as navigating Google Maps or searching for information.
Similar to OpenAI’s Operator, Open Computer Agent works by opening necessary programs and figuring out the required steps to complete a task. However, in TechCrunch’s testing, it struggled with more complicated requests like searching for flights and often encountered CAPTCHA tests it couldn’t solve.
Users may also experience wait times ranging from seconds to minutes due to a virtual queue, depending on demand. Aymeric Roucher, a member of Hugging Face’s agents team, said the goal was to demonstrate that open AI models are becoming more capable and cheaper to run on cloud infrastructure.
“As vision models become more capable, they become able to power complex agentic workflows,” Aymeric Roucher, stated in a post on X. “[Some of these models] support built-in grounding, i.e. [the] ability to locate any element in an image by its coordinates, [and] thus [can] click any item [in a virtual machine].”
Roucher noted that advancements in vision models, such as Qwen-VL, enable complex agentic workflows by supporting built-in grounding, allowing them to locate and interact with elements in an image. The release aims to showcase the potential of open AI models.
The development of agentic technology is attracting significant investment as enterprises look to boost productivity. A KPMG survey found that 65% of companies are experimenting with AI agents, and Markets and Markets projects that the AI agent segment will grow from $7.84 billion in 2025 to $52.62 billion by 2030.
