Composer 2.5: Pricing, Fast Mode and Uses

Composer 2.5The current Composer release, better at long-running tasks and at judging when a job needs a light touch versus deep work. Press Enter for the full definition. is Cursor's own coding model, trained for long tasks that use files, terminal commands, and other agent tools. Cursor rates it fast, low cost, and frontier intelligence. The release improved sustained work, complex instruction following, effort calibration, tool selection, intent understanding, communication, and reliability over Composer 2Cursor's in-house agentic coding model: frontier-level coding quality at high speed and low cost, built as a software-engineering specialist rather than a general-purpose model. Press Enter for the full definition..

It uses a 200k context window and has no larger maximum listed on the model card. That is smaller than the 1M ceilings on some third-party frontier models, but context size is only one part of task performance. ComposerCursor's own fast coding model, tuned for the editor and priced well below frontier models; the recommended day-to-day model for executing a plan. Press Enter for the full definition.'s value is the write, run, inspect, and correct loop inside Cursor, especially when a task needs many tool calls and the per-token cost matters.

Use Fast for an interactive pairing loop where response time keeps a person moving. Use Standard for queued cloud work, high-volume batches, or any run where nobody is waiting on each tool result. Cursor says the two variants have the same intelligence. The difference is speed and price, and Fast is selected by default in the product.

Fast costs six times the Standard input and output rates, with cache reads at $0.50 instead of $0.20. That premium can make sense for a short feedback loop and looks wasteful on a background migration that will take time either way. Check the selected variant explicitly because the plain model name can resolve to the product default in some surfaces.

Do one comparison with the same task before changing a team default. Record elapsed time, usage, final checks, and human waiting time. Fast can be worth more than its token premium if it keeps a high-cost review session moving, while Standard can win easily when an agent runs unattended overnight.

Cursor describes a substantial improvement in intelligence and behavior. The model stays with long-running work more reliably, follows complex instructions better, calibrates effort to task difficulty, selects tools more carefully, and communicates more clearly. Those behavior changes are valuable because a long task can fail through one confused tool call or a lost constraint even when its benchmark score looks strong.

The research release explains part of the training change. Composer 2.5The current Composer release, better at long-running tasks and at judging when a job needs a light touch versus deep work. Press Enter for the full definition. remained on Moonshot's Kimi K2.5The open base model Cursor continued-trained into Composer 2 (1T parameters, 32B active, 256K context), chosen mainly for how well it fit Cursor's serving infrastructure. Press Enter for the full definition. open checkpoint, then Cursor scaled its training, built harder reinforcement-learning environments, used targeted textual feedback around specific mistakes, and generated 25 times more synthetic tasks than Composer 2Cursor's in-house agentic coding model: frontier-level coding quality at high speed and low cost, built as a software-engineering specialist rather than a general-purpose model. Press Enter for the full definition.. Cursor notes that benchmark scores do not capture all of the communication and effort-calibration work.

ComposerCursor's own fast coding model, tuned for the editor and priced well below frontier models; the recommended day-to-day model for executing a plan. Press Enter for the full definition. is a strong starting point for everyday implementation, repetitive repository work, and long agent runs where tool use matters more than broad world knowledge. It can search the project, edit files, run the terminal, use the browser, search the web, generate images, ask questions, and fetch rules under the current surface's permissions.

A cheap model that needs three retries is not cheap, and an expensive model that adds no outcome is not better. Keep a small evaluation set from real work and compare cost per accepted task. ComposerCursor's own fast coding model, tuned for the editor and priced well below frontier models; the recommended day-to-day model for executing a plan. Press Enter for the full definition.'s published pricing makes it attractive; the evaluation tells you where that advantage survives your quality bar.

Composer 2.5The current Composer release, better at long-running tasks and at judging when a job needs a light touch versus deep work. Press Enter for the full definition. draws from the Cursor ModelsCursor's plan and usage pool for models it labels as Cursor Models, kept separate from the third-party Other Models pool. Press Enter for the full definition. pool with Grok 4.6A Cursor and SpaceXAI frontier model for complex coding and knowledge work; it is a selectable model and is separate from the Grok Bot product. Press Enter for the full definition. and Grok 4.5. Cursor describes that pool as carrying significantly more included usage than the third-party pool. It is available across Cursor's desktop, web, CLI, and mobile surfaces, and it can be used through the Cursor SDK where the plan supports that product.

Cursor StartAn India-only individual plan centered on Cursor Models, with cloud and mobile access but without Other Models, Auto, Bugbot, Automations or the Agent SDK. Press Enter for the full definition. includes Composer 2.5The current Composer release, better at long-running tasks and at judging when a job needs a light touch versus deep work. Press Enter for the full definition. in Standard mode and does not permit Fast. Pro and higher plans expose the broader controls according to their current terms. On Teams and Enterprise, models in the Cursor ModelsCursor's plan and usage pool for models it labels as Cursor Models, kept separate from the third-party Other Models pool. Press Enter for the full definition. pool are exempt from the separate Cursor Token Rate that can apply to directly selected third-party models and some Auto routes.

Use the dashboard pool balance rather than translating the list rate into an assumed request allowance. Each task consumes a different amount of input, cached context, output, and tool-loop history. The plan tells you what pool is available; the dashboard shows how your actual workload spends it.

ComposerCursor's own fast coding model, tuned for the editor and priced well below frontier models; the recommended day-to-day model for executing a plan. Press Enter for the full definition. is also the useful baseline when evaluating a new frontier model. Run the cheaper Cursor model first on a task you can verify, then escalate only if the result misses. This keeps the comparison tied to work the team really does and stops a launch benchmark from deciding every model selection in advance.

For shared workflows, record the variant as well as the model. A report that says Composer 2.5The current Composer release, better at long-running tasks and at judging when a job needs a light touch versus deep work. Press Enter for the full definition. without Standard or Fast leaves out the largest price difference in the choice. Add the surface too, since product defaults and explicit SDK parameters can resolve the same short model ID differently.

Review that baseline after meaningful model or harness changes. Cursor can improve the same model ID, and repository tasks change as the codebase grows. Repeating a small, stable evaluation set gives the team current evidence without turning model selection into a permanent testing project.