# Cursor 3

> Cursor 3, out April 2, rebuilds the editor around orchestrating agents rather than typing every line. Its headline feature lets you launch up to eight AI agents in parallel, each working autonomously on its own isolated Git branch, so you review and merge their output instead of babysitting a single assistant. On standard implementation tasks the company measured 30 to 50 percent speed gains. Under the hood is Composer 2, Cursor's own frontier coding model, paired with a multi-repo layout and smooth handoff between local and cloud runs. It is a direct answer to Claude Code and OpenAI's Codex, and a clear bet that the near future of programming looks less like writing code and more like managing a small team of agents that do.

_Section: [New AI Tools](https://www.wortins.com/new-tools) · Source: Cursor · Published Thursday, July 9, 2026_

## Wortins' read

Cursor 3, out April 2, rebuilds the editor around orchestrating agents rather than typing every line. Its headline feature lets you launch up to eight AI agents in parallel, each working autonomously on its own isolated Git branch, so you review and merge their output instead of babysitting a single assistant. On standard implementation tasks the company measured 30 to 50 percent speed gains. Under the hood is Composer 2, Cursor's own frontier coding model, paired with a multi-repo layout and smooth handoff between local and cloud runs. It is a direct answer to Claude Code and OpenAI's Codex, and a clear bet that the near future of programming looks less like writing code and more like managing a small team of agents that do.

## Source

[Read the full story at Cursor](https://cursor.com/blog/cursor-3)

## Related coverage

- [Agentjacking Attack Tricks AI Coding Agents Into Running Malicious Code](https://www.wortins.com/story/agentjacking-attack-tricks-ai-coding-agents-into-running-mal-be967ff8) — [The Hacker News](https://thehackernews.com/2026/06/agentjacking-attack-tricks-ai-coding.html)
- [Taktile Raises $110 Million Series C](https://www.wortins.com/story/taktile-raises-110-million-series-c-2e39ee5e) — [Taktile](https://blog.mean.ceo/ai-startup-funding-news-july-2026/)
- [METR Detects Highest Benchmark Gaming Rate in GPT-5.6 Sol's Evaluation](https://www.wortins.com/story/metr-detects-highest-benchmark-gaming-rate-in-gpt-5-6-sol-s--1dcc8649) — [TechTimes](https://www.techtimes.com/articles/319662/20260703/ai-benchmark-cheating-sets-record-gpt-56-sol-gamed-its-own-safety-tests.htm)
- [Codex Micro](https://www.wortins.com/story/codex-micro-3d90f9e1) — [DevOps.com](https://devops.com/openai-expands-into-developer-hardware-with-codex-micro-keyboard/)
- [xAI Releases Grok 4.5, Its Largest Model at 1.5 Trillion Parameters](https://www.wortins.com/story/xai-releases-grok-4-5-its-largest-model-at-1-5-trillion-para-be0c5aaf) — [xAI](https://x.ai/news)
- [Amazon One Medical](https://www.wortins.com/story/amazon-one-medical-8d562045) — [FierceHealthcare](https://www.fiercehealthcare.com/ai-and-machine-learning/amazon-one-medical-releases-agentic-health-ai-assistant-members)

---

_Curated and written by [Wortins](https://www.wortins.com) — The daily AI briefing. Every story links to its original source; the "Wortins read" on each is our own original analysis. [About Wortins & our editorial approach](https://www.wortins.com/about)._