Cursor

AI-native code editor: the gold standard for Tab completion + in-house Composer models, currently being acquired by SpaceX in a $60 billion deal.

Cursor is the benchmark among AI-native code editors—forked from VS Code, with AI welded deep into every layer of the editor. Tab completion is powered by the in-house Fusion model and delivers the best “feel” in its class; version 3.0 (April 2026) shifted to an agent-first interface where up to 8 agents work in parallel on separate git worktrees, and with cloud agents plus an iOS app that launched in June, you can even run your coding agents from your phone. The in-house model lineup has also come into its own: Composer leads with “4x the speed at equal intelligence,” Composer 2.5 was post-trained on Kimi K2.5’s open-source weights, and the next generation is being trained from scratch in partnership with SpaceX, using compute equivalent to one million H100s.

Pricing: the free Hobby tier comes with limited quota; the workhorse is Pro at $20/month, with Pro+ at $60 and Ultra at $200 (20x quota) above it, and Teams at $40/user/month. Commercially, it’s the most powerful rocket in the industry: ARR crossed $1 billion in November 2025 at a $29.3 billion valuation; in June 2026, SpaceX announced an all-stock acquisition of parent company Anysphere for $60 billion, expected to close in Q3—the largest M&A deal to date in the AI coding space.

Ecosystem niche: the IDE-native form factor (versus Claude Code’s terminal-based form and Copilot’s plugin form), with core selling points of free multi-model switching and the deepest editor integration—Tab, multi-file agents, Bugbot code review, MCP, and cloud agents, all in the box. It suits a “write-and-refine-as-you-go” workflow, and many people pair it with Claude Code: writing day-to-day in Cursor while handing off entire tasks to Claude Code.

Caveats: pricing rules have changed multiple times and quotas are opaque—heavy use can lead to “bill shock,” so high-volume users should keep a close eye on the usage dashboard; Privacy Mode is off by default, meaning your code may be used for model improvement, so be sure to enable it manually for commercial projects; and the proprietary fork means migrating back to vanilla VS Code or JetBrains means losing the entire AI layer—there’s some lock-in risk.

Source: Agentopia · Site-wide machine catalog available at /llms.txt