Cursor Launches Agent-First Coding Product to Compete With Claude Code and Codex as AI Labs Encroach

Cursor has launched Cursor 3 (codenamed Glass), a major product overhaul that shifts the AI coding startup from an IDE-centric model to an agent-first approach, directly competing with Anthropic's Claude Code and OpenAI's Codex.

The new interface places a natural language task box at center stage, allowing developers to assign entire coding tasks to AI agents that work autonomously. Unlike Claude Code and Codex, Cursor 3 integrates this agentic approach within its existing AI-powered development environment, letting developers review AI-generated code locally while agents work in the cloud.

"In the last few months, our profession has completely changed," said Jonas Nelle, one of Cursor's heads of engineering. "A lot of the product that got Cursor here is not as important going forward anymore."

The launch comes as Cursor faces existential pressure from AI labs. Multiple developers told WIRED they have shifted from Cursor to Claude Code and Codex, largely driven by heavily subsidized subscriptions offering over $1,000 worth of usage for $200/month plans.

Cursor, reportedly raising fresh capital at a $50 billion valuation, is responding with in-house AI models. The startup recently launched Composer 2, based on Moonshot AI's open-source architecture with additional pretraining and post-training. The company says it has other strategies beyond competing on price.

The AI coding market has become one of the most competitive segments in the industry, with OpenAI, Anthropic, Cursor, and others vying for developers who increasingly rely on AI agents to write code rather than writing it themselves.