Anthropic has launched a web app for Claude Code. The new interface lets developers create and manage coding agents from a browser. The web version is available now to Pro and Max subscribers.
Web launch details
The web app sits at claude.ai and appears as a Code tab for eligible users. Pro subscribers pay twenty dollars per month. Max subscribers pay one hundred or two hundred dollars per month, depending on the tier. Anthropic said the web version runs on Anthropic-managed cloud infrastructure and supports parallel coding tasks. The company positions this build as a complement to the command line interface and a way to meet developers where they work.
Anthropic said the web rollout is a research preview. The company said the terminal experience will remain the most powerful and most customizable option for power users. At the same time, Anthropic said it aims to put Claude Code into the browser and into mobile apps so developers can use agents from more places. Product managers told reporters they want the CLI to stay central while making the product easier to access.
Why this matters
Claude Code has become a major revenue driver for Anthropic. The company says the product now represents a large slice of revenue and that usage has grown rapidly since the broader launch in May. Third-party reporting estimates the annualized revenue from Claude Code at over $500 million, and user counts have risen sharply. That scale explains why Anthropic is expanding access beyond terminals.
The web release also reflects how coding tools are changing the workflow. Early tools worked like autocomplete. New tools act as agents. Developers can spin up multiple agents that run in parallel. That shift turns many engineers into managers of AI assistants. Anthropic executives say a large part of their internal code base is now produced by Claude. The change is real at Anthropic and visible in how teams work today.
What developers should weigh
The new web app makes Claude Code easier to reach. That can speed up small tasks and routine fixes. It can also raise questions on accuracy and cost. Researchers have found that AI tools can help and can also slow some engineers down because of the time needed to craft prompts and to vet results. Teams should set clear checks so that agents do not introduce errors into critical projects. It will be revealed where the agents can be assisted by benchmarks and thorough testing.
Competition is intense. Code tools are also run by Microsoft, Google, OpenAI, Cursor, and others, both on the web and within editors. Anthropic now competes directly in that space. The web app and mobile access give Claude Code a broader footprint. The terminal will likely remain the place for complex workflows, while the web app will suit quick tasks and parallel work. Developers should try both interfaces and choose the flow that matches their code base and their safety needs.