The headline feature enables Cursor's main model to spin up independent subagents that run concurrently with their own isolated contexts, tools, and memory.

photograph, illustration, 3D render, or AI-generated artwork and generate up to 10 seconds of high-fidelity, cinematic motion while preserving character consistency, lighting continuity, and artistic style.

connect personal health information from Apple Health (iOS), Health Connect (Android), HealthEx, and Function Health.

The models support translation across 55 languages and are designed to deliver high translation quality while maintaining strong efficiency for on-device and low-resource deployments.

securely connects information from Gmail, Google Photos, Search history, and YouTube watch history to deliver more personalized and context-aware responses.

to generate eight-second clips from up to three reference images combined with a text prompt.

accurate pronunciation across multiple languages, and lifelike expressiveness suitable for real-world deployment. The model supports zero-shot voice cloning, fine-grained control over emotion and speaking style, and generation of natural-sounding speech in diverse languages and accents.

to everyday non-technical tasks across documents, emails, spreadsheets, presentations, and project management.

loyalty programs, and secure checkout using JSON manifests that merchants publish to expose structured commerce data.