4x Faster Coding with AI? Meet Composer by Cursor

Summary

Cursor unveiled its 2.0 platform alongside a new AI model called Composer, which the company says generates code up to four times faster than competing models while delivering near‑state‑of‑the‑art quality. Composer appears to be a fine‑tuned version of a Chinese open‑source code model, enhanced with reinforcement learning to handle large codebases and integrate coding tools. Early internal benchmarks and external code‑related tests show the model performing strongly against rivals such as Entropic, positioning it as a fresh alternative for developers using Cursor. The rollout invites existing users and the broader developer community to evaluate its speed and accuracy gains.

Original Description

🚀 Cursor just dropped Composer, their new frontier coding model—and it’s wild. It hits 4x faster generation on their benchmarks while keeping frontier-level coding performance.

The secret? They trained it on real-world software engineering tasks, not toy examples—then reinforced it to act like a real dev inside complex codebases. Composer can read/edit files, run terminal commands, and even perform semantic code searches.

What’s cooler: since response speed is key for live coding, they trained it to optimize tool use and parallelism—so you get not just smart output, but instant feedback.

Definitely one to watch if you care about developer-focused AI models. ⚙️

I’m Louis-François, PhD dropout, now CTO & co-founder at Towards AI. Follow me for tomorrow’s no-BS AI roundup 🚀

#AInews #LLM #cursor #short