Cursor releases Composer 2.5 for improved AI coding assistance
Written byMango
Drafted with AI; edited and reviewed by a human.
![]()
TL;DR
- Cursor has launched Composer 2.5, an advanced AI model for its coding assistant.
- This new version offers significant improvements in intelligence and behavior over its predecessor, Composer 2.
- Composer 2.5 excels at long-running tasks and following complex instructions, making collaboration more effective.
- Special introductory offer includes double usage for the first week.
Cursor has just unveiled Composer 2.5, a major upgrade to its AI coding assistant. This new iteration promises a substantial leap forward in both the intelligence and overall behavior of the tool compared to the previous Composer 2 model. Developers can expect a more capable AI partner that understands and executes complex coding tasks with greater precision and reliability.
One of the standout improvements in Composer 2.5 is its enhanced ability to handle sustained work on long-running tasks. This means that for more involved projects or intricate debugging sessions, the AI will maintain context and coherence, providing more consistent and useful assistance throughout the process. This is a critical development for tackling large codebases or multi-stage development cycles.
Furthermore, Composer 2.5 has been engineered to follow complex instructions more reliably. Developers often provide multi-faceted prompts or specific constraints when seeking AI assistance. This enhanced adherence to detail ensures that the AI's output is more aligned with the developer's precise requirements, reducing the need for extensive revisions or clarifications.
Collaboration with the AI has also been a focus, with Composer 2.5 described as being more pleasant to collaborate with. This suggests improvements in the AI's conversational flow, its ability to ask clarifying questions, and its overall responsiveness, creating a more seamless and productive interaction for developers. You can explore the full technical details in the Model Documentation.
The pricing for Composer 2.5 is structured into two tiers: Standard and Fast. The Standard tier is priced at $0.50 per million input tokens and $2.50 per million output tokens. The Fast tier, which is the default option, comes at a higher rate of $3.00 per million input tokens and $15.00 per million output tokens, reflecting its enhanced performance and speed.
As an incentive to try the new model, Cursor is offering double usage for the first week of Composer 2.5 availability. This allows developers to leverage the enhanced capabilities more extensively during the initial rollout period, providing ample opportunity to experience the benefits firsthand. For more information on the release, check out the Announcement.
Summary
- Composer 2.5 is now available in Cursor, bringing enhanced AI capabilities to the coding assistant.
- Key improvements include better handling of long-running tasks and more reliable execution of complex instructions.
- Pricing details are available, with a special introductory offer of double usage for the first week.
- Further information and technical specifications can be found on the Composer 2.5 Changelog.
Source: Composer 2.5 · Cursor
Read next

Claude Managed Agents Gain Self-Hosted Sandboxes and MCP Tunnels for Enhanced Security
Anthropic introduces self-hosted sandboxes and MCP tunnels for Claude Managed Agents, allowing users to run agent execution within their own infrastructure for increased security and control.
Continue reading