Introducing Cursor 2.0. Our first coding model and the best way to code with agents.

I spent the last 60 days working at Cursor. It's been one of the most thrilling phases of my professional life. There's a lot of mystique around the company. Over the last two months, some things matched my expectations; many did not. I wrote an essay for @joincolossus about things that have surprised me about the company and its culture so far. joincolossus.com/article/ins…

Semantic search improves our agent's accuracy across all frontier models, especially in large codebases where grep alone falls short. Learn more about our results and how we trained an embedding model for retrieving code.

Finally, we've drastically improved the performance of using LSPs. Python and TypeScript LSPs are now faster by default. Memory is dynamically configured based on available RAM. We've also fixed a number of memory leaks and improved memory usage.

We've updated our web interface.

A new confirmation modal to prevent accidentally closing Cursor.

Cloud agents are now available in-editor. Switch from local → cloud to run agents while you're away from your computer.

We've simplified the prompt input.

Review the agent's changes across your codebase without needing to jump between individual files.

We've improved the agent harness for all models.

You can now plan with one model and implement with another. Build the plan locally or in the cloud.

We've shipped several quality-of-life improvements to Cursor! Details below...

Here’s everything new and how our team uses cloud agents: cursor.com/blog/cloud-agents

Manage a fleet of cloud agents, directly from Cursor. Now with faster startup, improved reliability, and a new UI.

Composer is a new model we built at Cursor. We used RL to train a big MoE model to be really good at real-world coding, and also very fast. cursor.com/blog/composer Excited for the potential of building specialized models to help in critical domains.

Read about everything that's new in 2.0. More tomorrow! cursor.com/blog/2-0

Turn speech into code with voice mode.

With a built-in browser, agents can now run and test their code.

Easily run the same prompt across models for maximum intelligence.

Composer is a frontier coding model that completes tasks in under 30 seconds.