Cursor, the AI coding platform from San Francisco startup Anysphere valued at $29.3 billion, has launched Composer 2, a new in-house coding model integrated directly into its development environment. The company is positioning the model as a significant upgrade for long-horizon agentic coding tasks rather than just improved code completions.
The pricing represents a dramatic cost reduction from previous versions. Composer 2 Standard costs $0.50 per million input tokens and $2.50 per million output tokens, while the faster variant costs $1.50 and $7.50 respectively. This marks an 86% price drop from Composer 1.5, which launched in February at $3.50 per million input tokens and $17.50 per million output tokens. Cache-read pricing for repeated tokens is also reduced to $0.20 million for standard and $0.35 million for the fast version.
Unlike many AI model releases, Composer 2 appears to be exclusively available within Cursor's platform rather than as a standalone API or through external model providers. The company describes it as specifically tuned for Cursor's agent workflow and integrated with the product's existing tool stack, suggesting a tightly coupled approach to AI-assisted development.
The launch signals continued competition in AI coding tools, with companies like Cursor betting that specialized, integrated models will outperform general-purpose alternatives for specific workflows. The dramatic price cuts also indicate improving economics in AI model deployment, potentially pressuring competitors to reduce their own pricing structures.