Composer Models by Cursor (Anysphere)
2 models2026Up to 200K ctxFrom $0.5/1M input
About
Cursor's proprietary family of agentic coding models, designed for long-horizon software engineering inside the Cursor IDE. Each generation is trained with reinforcement learning on real engineering tasks and evaluated on CursorBench and Terminal-Bench. Generations: Composer (Oct 2025), Composer 1.5, Composer 2 (Mar 2026), Composer 2.5 (May 2026). All models are Cursor-native; not available as standalone third-party APIs.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
2 in view
Composer 2.5Current
Use when the workload needs code, tool use, and function calling.
2026-05codetool usefunction calling
Composer 2Current
Use when the workload needs code, 200K context, and tool use.
2026-03code200K contexttool use
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Composer 2.5 | Use when the workload needs code, tool use, and function calling. | 2026-05 | codetool usefunction calling | Current |
| Composer 2 | Use when the workload needs code, 200K context, and tool use. | 2026-03 | code200K contexttool use | Current |
Release Timeline
2 release groups2026-05
1 current
Composer 2.5
Currentcodetool usefunction calling
2026-03
1 current
Composer 2
Currentcode200K contexttool use
Specifications(2 models)
| Model | Released | Context | Fn Calling | Tool Use | Code Exec |
|---|---|---|---|---|---|
| Composer 2.5 | 2026-05 | — | Yes | Yes | Yes |
| Composer 2 | 2026-03 | 200K | Yes | Yes | Yes |
Available From(1 provider)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Composer 2.5 | Cursor | $0.5 | $2.5 | Serverless |
| Composer 2 | Cursor | $0.5 | $2.5 | Provisioned |
Comparisons
- Composer 2.5 vs Claude Opus 4.6
- Composer 2.5 vs Claude Sonnet 4.6
- Composer 2.5 vs GPT-5.4
- Composer 2.5 vs DeepSeek V3
Frequently Asked Questions
- What is Composer used for?
- Composer is used for code, agent workflows and tool use, and code execution. The family description and listed model capabilities point to those workloads as the best fit.
- How does Composer compare to Claude 3?
- Composer by Cursor (Anysphere) is strongest where you need code, while Claude 3 by Anthropic is the closest related family to check for vision and multimodal work. Composer has 2 listed variants and reaches up to 200K context, while Claude 3 reaches up to 200K context, so compare the specs and pricing tables before choosing a production model.
- Which Composer model should I use?
- For the lowest listed input price, start with Composer 2.5 through Cursor at $0.5/1M input tokens. For the most capable/latest local choice, evaluate Composer 2 with 200K context and tool use and function calling.