Long-horizon coding
Trained through reinforcement learning for software tasks that could require hundreds of agent actions.
Historical tool archive
Composer 2 was Cursor's in-house coding model released in March 2026 for long-horizon software tasks. It has since been superseded by Composer 2.5 and is no longer listed in Cursor's current model catalog.
No longer active

Archive
Cursor introduced Composer 2 on March 19, 2026 as a lower-cost agentic coding model trained for tasks requiring hundreds of actions. It improved substantially over Composer 1 and 1.5 on Cursor's published internal and external benchmark results.
Composer 2 had a standard tier priced at $0.50 per million input tokens and $2.50 per million output tokens, plus a faster tier at $1.50 input and $7.50 output. Cursor released Composer 2.5 on May 18, 2026 as a direct improvement in intelligence, long-running-task performance, instruction following, communication, and effort calibration.
Past capabilities
Trained through reinforcement learning for software tasks that could require hundreds of agent actions.
Used Cursor's first continued-pretraining run to strengthen the base before reinforcement learning.
Offered a lower-cost standard model and a higher-throughput fast option with the same stated intelligence.
Ran inside Cursor and drew from the included first-party model usage pool on individual plans.
Availability
Composer 2 is a historical model and is no longer included in Cursor's current model catalog. Its launch pricing was $0.50 per million input tokens and $2.50 per million output tokens; the fast variant was $1.50 input and $7.50 output. Cursor's current Composer option is Composer 2.5.
Status checked . Review the historical source ↗
Current options
Coding
Composer 2.5 is the direct current successor and keeps the standard $0.50/M input and $2.50/M output pricing.
Explore Composer 2.5 →Coding
A current coding-agent alternative for terminal-driven repository work and long-running engineering tasks.
Explore Claude Code →Coding
A current agentic coding platform for parallel tasks, repository changes, review, and cloud execution.
Explore Codex →Questions
Composer 2 is no longer listed in Cursor's current model catalog. Composer 2.5 is the current direct successor.
Cursor released Composer 2.5 in May 2026 with better long-running-task performance, instruction following, communication, and effort calibration.
At launch, the standard model was $0.50 per million input tokens and $2.50 per million output tokens. The fast variant was $1.50 input and $7.50 output.

Replit Agent 4 - Replit's parallel AI coding agent with infinite design canvas and task-based team workflows

Cursor 3 - Cursor's agent-first interface for running parallel coding agents across repos

Code Review - Anthropic's multi-agent PR review system for Claude Code

SWE-1.6 - Cognition's updated coding model for Windsurf optimized for speed (950 tok/s) and smoother agent UX

Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.