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

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

Easily run the same prompt across models for maximum intelligence.

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

Turn speech into code with voice mode.

Read about everything that's new in 2.0. More tomorrow!

brb,

I would like the brain icon to appear in this dropdown preview when I've selected the Thinking version, pls

holy shit!

We would love to benchmark this using coding benchmarks that we run like-for-like across all models. Let us know!

I’m sorry, what do you mean multiple models at the same time? Who’s gonna pay for the tokens? Did you also dropped the prices?

For existing users, switch to Agent mode in Cursor Settings

I've been using this for the past week and let me tell you, it's amazing! The new model, the improved browser integration, and overall the fit-and-finish of the product is pretty great. Awesome job on this.

First test drive. Its fast, then it deleted the entire file without telling me . great

Huge upgrade! 👀👀👀

Context engineering in Cursor spikes high up in Cursor now because its Composer gives it power with awareness and project-level intelligence. To clarify, if you build MVP, with Composer you can be more confident it’s gonna handle the whole scope of the project

tl;dr for 2.0 launch: - composer model (a fast code-gen model): 4x faster than similarly intelligent models, finishing in under 30 seconds - multi-agent interface: a new interface designed focus on agents, not files. developers can run many agents in parallel to compare there’s also a built-in browser tool that lets cursor natively test and iterate its own solutions

Congrats on the launch!

what is the goal of executing multiple agents with the same task? do they interact with each other? or is there a third one reviewing and picking the best option? i dont get it yet :/

excited to use and criticize heavily!

So fast, won't let me doom scroll on Twitter at all. Congratulations on the release!

The day has finally come! Here for the voice and browser modes especially - LFG team! ✨

Y’all cooking. I had the opportunity to test Composer 1 on a difficult task and it performed way better than all the LLMs I tried.

I‘ve been using @cursor_ai 2.0 in the early access program and I love it! Favorite features: - Agent code review - Voice input - Integrated browser - with ability to select HTML elements as context - Multi model tasks - give same task to different models and compare

Wait, so does Composer use tokens from all three models? Isn’t that quite wasteful?

Read our full vibe check on the new release @every

yo what

I just spun up 4.5 sonnet and codex in parallel and both of them spent thousands of tokens trying to navigate to a python script to change the color of a chart

bro cursor 2.0 just dropped n my laptop started sweating 💀 agents coding agents while im just vibing here wtf did we even do before this lmao

built in browser feature is sick

What is the pricing for the composer model?

🤯

only app people love updating!

ok cursor, build me a million dollar AWS alternative, make no mistakes 🌝

Delighted to have had early access to Cursor 2.0 for a week or so. New features are a game changer. Really a worth-while major version release. Well-done Cursor team!

This is cool stuff but we need some tutorials and the docs are lacking. Why would we run multiple agents via Worktree? How is that different than "Local" agent with a 2x "cycle count"? What is a "cycle count"?

whatever you build with cursor, all you need is reddit marketing 👇

Let’s go!

Congrats to the whole team on the launch 👏 I had a lot of fun testing Cursor 2.0. and who knows, maybe one day I'll have more than a guest account.

Bang