Cursor Launches Composer 2 AI Coding Model
Anysphere has introduced Composer 2, a new proprietary AI coding model developed for its Cursor platform. Through this release, the company aims to advance autonomous coding capabilities while competing with OpenAI and Anthropic.
At the same time, the launch coincides with reports that the company is exploring a new funding round at a valuation of around $50 billion. As a result, Composer 2 arrives during a period of rapid growth and heightened investor interest.
Performance gains and pricing strategy
Composer 2 represents the third iteration of Cursor’s in-house models since October 2025. Moreover, it builds on earlier versions with improved performance on coding benchmarks such as Terminal-Bench 2.0 and SWE-bench Multilingual. Consequently, the model handles long, multi-step coding tasks that require hundreds of sequential actions.
In addition, the model uses reinforcement learning trained on extended coding workflows. Therefore, it can perform more complex, sustained development tasks with higher accuracy. Furthermore, Cursor attributes these improvements to a continued pretraining phase, which strengthens the model’s base capabilities.
Pricing also reflects a competitive positioning. The model costs $0.50 per million input tokens and $2.50 per million output tokens. Meanwhile, a faster variant is available at higher rates, offering flexibility based on performance requirements.
Rapid adoption and market expansion
Alongside technical progress, Cursor has seen strong user growth. The platform now records over one million daily active users and more than seven million monthly active users. In addition, enterprise adoption continues to rise, with large organizations deploying the tool across thousands of developers.
Revenue growth has also accelerated. Cursor’s annualized revenue exceeded $2 billion in February, while corporate customers now contribute a majority share. As a result, the company has strengthened its position within the enterprise AI tooling segment.
Intensifying competition in AI coding tools
However, competition in the AI coding space continues to intensify. Microsoft maintains strong traction with GitHub Copilot, while OpenAI expands its Codex ecosystem. At the same time, Anthropic advances its Claude Code offering, and Google has introduced new AI development tools such as Antigravity.
Therefore, Cursor faces increasing pressure from well-funded competitors with extensive resources. Nevertheless, Composer 2 represents a key step in strengthening its proprietary model strategy.
Ultimately, the company’s future growth depends on its ability to sustain its performance gains as it scales its platform. As competition accelerates, continued innovation will remain critical to maintaining its position in the evolving AI coding landscape.
