AI code editor Cursor has surpassed GitHub Copilot in active users.


Cursor just passed GitHub Copilot in active users. Let that sink in — a startup beat a Microsoft-backed product in developer tools, a category Microsoft has dominated for decades. Surprising? Not if you've actually used both.
Two words: Agent Mode. This isn't autocomplete that finishes your lines of code. Cursor's agent edits files across your project, runs tests, fixes bugs, creates pull requests — autonomously. Think of it as a junior developer who works around the clock and never opens LinkedIn during sprints.
The migration from VS Code is frictionless because Cursor is literally built on top of it. Extensions, keybindings, themes — everything carries over. But the AI layer is on another planet. The model sees your entire codebase, not just the open file. It understands project architecture, module dependencies, and coding conventions. GitHub Copilot autocompletes lines; Cursor understands systems.
Multi-model support is another edge. GPT-5, Claude, Gemini — switch depending on the task. In our workflow, Claude handles refactoring best, GPT-5 is stronger for greenfield code, and Gemini excels at documentation tasks. Copilot locks you into one model. That rigidity costs you.
Background Agents let you kick off large refactoring or migration jobs and keep working on other things. In practice, we've found it works best when you break big tasks into smaller chunks rather than letting it loose on an entire codebase at once.

5 million active developers. 120% growth quarter over quarter. The company says a significant chunk of new users are former GitHub Copilot subscribers switching over. Average time saved: 2-3 hours per day per developer.
Take those numbers with a healthy dose of skepticism — every company defines "active users" differently. But the direction is unmistakable: the agent paradigm is eating autocomplete alive.
GitHub Copilot still has the best autocomplete in the business and wins on GitHub ecosystem integration — PRs, issues, Actions. If your entire workflow lives inside GitHub, that's a real advantage. Windsurf (the tool formerly known as Codeium) offers agent features at a lower price point but can't match Cursor's quality yet. JetBrains AI Assistant is evolving slowly — JetBrains seems to be betting on caution over speed.
Cursor works globally, no restrictions. Pro is $20/month, Business is $40/month. There's a free tier with limited AI requests — enough to see what the fuss is about before committing. If you write code daily, the Pro subscription pays for itself within the first week. We ran the numbers on our own team: 2 hours saved per day at even a modest hourly rate makes $20/month look like a rounding error.
OpenAI has unveiled GPT-5 — its most powerful language model yet. We break down the key improvements and differences from GPT-4.
ReadMidjourney has launched a web interface with an image editor, accessible without Discord.
ReadSuno has released version 4 — music generation has reached studio quality.
Read