AI-first IDE forked from VS Code, by Anysphere. Tab completion is the headline (deep multi-file edit predictions), plus Composer for repo-wide changes, agent mode, and chat that understands the whole codebase via embeddings. Bring-your-own model (Claude, GPT-5, Gemini). Privacy mode keeps your code out of training.
No pricing data on file.
Built as a fork of VS Code with AI integration baked into the editor itself rather than bolted on as an extension. The differentiator is Composer/Agent mode, which reads across multiple files, proposes multi-file edits, runs terminal commands, and iterates on test failures inside a single conversational loop. Codebase indexing is the other quiet superpower — the editor builds a semantic index of the repo so prompts can reference symbols and files without manual context-pasting. Tab completion runs on Cursor's own model and is fast enough that most users leave it on by default.
Best fit for engineers shipping greenfield code or doing focused refactors on repos under a few hundred thousand lines. Solo developers and small teams adopt it most aggressively; larger enterprises bump into the privacy-mode conversation and the lack of a real on-prem story.
Watch out for: the premium request tier still routes to Anthropic and OpenAI APIs, so cost and latency follow upstream weather. Repo indexing degrades noticeably on multi-million-line monorepos, and Agent mode will confidently rewrite working code if you let it run unattended on a task it has misread.