Large language model developer Anthropic PBC today rolled out an update for Claude Code, an agentic artificial intelligence software tool for developers that helps turn ideas into code, upgrading it with a metrics dashboard for administrators.

Claude Code launched into general availability alongside the launch of Claude 4, the company’s flagship frontier AI model family, including Opus 4 and Sonnet 4. Opus is the company’s most powerful model yet, designed to sustain the performance of complex, long-running tasks, such as those that might take thousands of steps. Sonnet is a smaller, more generalized instruction-following model designed for speed and cost efficiency across high-volume tasks.

Enterprise admins will now have visibility into what developer teams are working on in real-time and how they are using AI tools. This includes lines of code accepted, tool usage rates, user activity breakdowns and cost tracking per user.

Claude Code runs in the terminal, which differs from a chat window or a development environment, in that it is where developers go to access the operating system directly.

The terminal is a text-based interface that allows users to interact with the computer by typing commands, providing a command-line interface for accessing applications and tools. This is separate from the graphical interface in Windows or macOS, where most interactions take place using icons and drag-and-drop. It’s best recognized as the black-and-white screen that people might remember from ‘90s hacker movies, where people type furiously at keyboards to get computers to perform magical tasks.

Anthropic said that since the launch of Claude 4, Clade Code’s active user base has grown by over 300% and its run-rate revenue has increased by more than 5.5 times. Public-facing companies using Code include Japanese technology conglomerate Rakuten Group Inc., collaborative web design tool Figma Inc. and communication software company Intercom Inc.

This news comes a month after the company announced model context protocol support for Code, which builds in the ability to connect external tools and services. MCP is an open standard pioneered by Anthropic that defines how AI models can securely access and use context from external tools, data sources and other resources. This allows Code to connect to tools such as a Figma server, OpenAI’s ChatGPT or image generation services, an external database from Oracle Corp. or numerous other sources.

On Monday, the company also announced that Claude Code is now natively available for Windows operating systems.

Claude Code follows a trend of new AI coding tools that work in the terminal, including from Google LLC and OpenAI, with releases of Gemini CLI and Codex CLI, respectively. According to a report from TechCrunch, these terminal-based AI products are already among the most popular among developers.

“Our big bet is that there’s a future in which 95% of LLM-computer interaction is through a terminal-like interface,” said Mike Merrill, co-creator of the leading terminal-focused benchmark Terminal-Bench.

Images: Anthropic