Contains affiliate links. We may earn a commission — it never changes the verdict.Details
The Stack / Coding
Cursor
03 CreationYes for anyone coding weekly — Cursor remains the best AI-in-the-editor experience, and Pro at $20 buys more agent work than it did a month ago: limits doubled across every plan on July 21, 2026. The pricing story changed shape — plans now sell agent-request limits against a first-party model pool (Grok 4.5 and Composer 2.5) rather than a dollar credit balance. Heavy agent sessions still climb the ladder; that's how $20 becomes $60.
The job in the stack
The hands-on coding seat: an editor where AI agents write, refactor, and navigate the codebase with you.
Pricing
| Plan | Price | Notes |
|---|---|---|
| Hobby | $0 | 2,000 completions/mo, 50 slow premium requests |
| Pro | $20/mo | Extended agent limits, generous Grok 4.5 + Composer usage, frontier model access, cloud agents, MCP/skills/hooks. Bugbot billed by usage |
| Pro+ | $60/mo | 3x Pro agent limits |
| Ultra | $200/mo | 20x Pro agent limits, priority access to new features |
| Teams Standard | $40/user/mo | Central billing, SSO, Bugbot code reviews, shared cloud-agent context. Teams Premium adds 5x agent limits |
Pricing verified 2026-08-30. Plans change — the source of truth is Cursor's own pricing page.
Models & versions
- Grok 4.5Released July 8, 2026 — a mixture-of-experts frontier model Cursor trained jointly with SpaceXAI on Cursor data and STEM tasks. Included in the first-party pool on paid plans; API rates $2/M in, $6/M out ($4/$18 fast)
- Composer 2.5Cursor's smaller first-party coding model — a different weight class from Grok 4.5, and Cursor says models of this size keep shipping
- AgentMulti-file agentic editing with frontier models (Claude, GPT, Gemini) alongside the first-party pool
The honest read
What works
- +Best-in-class tab completion — the daily 90% of AI coding value
- +Agent mode handles multi-file changes with reviewable diffs
- +Model-agnostic: route to Claude, GPT, or Gemini per task
What breaks
- −Usage credits on agent work make heavy months unpredictable
- −Pricing structure has churned repeatedly — expect it to move again
- −VS Code fork means occasional extension lag behind upstream
- −The SpaceX acquisition closed August 14, 2026 — Cursor is now a wholly owned subsidiary under SpaceX's new SpaceXAI division. The trust boundary is no longer hypothetical: check your privacy mode settings before pointing it at a sensitive codebase
Use it for
- →Developers living in an editor daily
- →Operators building internal tools who want AI pair-programming
- →Teams standardizing one AI coding environment
Skip it if
- ×You work terminal-first — Claude Code may fit tighter
- ×You code occasionally — Copilot's $10 plan is the cheaper on-ramp
Questions operators ask
?Cursor or GitHub Copilot?
Cursor if the editor is your home and you want agentic multi-file work. Copilot if you want cheap, unlimited completions inside stock VS Code. Full comparison on the compare page.
?Is Cursor Pro's $20 enough?
For completion-led work, easily — and July 21's across-the-board limit doubling bought Pro more room. If you run agents on large refactors daily, you'll still hit the agent-request ceiling; that's the signal to test Pro+, which is 3x Pro.
What changed
- News scan: the SpaceX acquisition closed August 14, 2026 (all-stock, $60B) — Cursor now operates as a subsidiary under SpaceX's new SpaceXAI division with access to the Colossus supercomputer; a Grok Bot beta (persistent, always-on agent teammates) has shipped. Separately, Cursor launched Origin on August 18 — a first-party code-hosting platform (repos, PR review, browser editing) positioned against GitHub amid GitHub's own reliability issues. Cons rewritten: the acquisition is done, not pending. Pricing ladder unchanged, still last verified Aug 2.
- News scan: xAI's Grok Bot (persistent, always-on agent teammates with browser/terminal access) is now bundled into Cursor Pro+, Ultra, and Teams seats at no extra cost, confirmed on cursor.com/help/grok-bot/plans — a direct product of the Cursor/SpaceXAI relationship. Noted here; doesn't change the verdict or pricing ladder, which stays unverified since the Aug 2 sweep.
- Real drift on cursor.com/pricing: the plan ladder holds ($0/$20/$60/$200, Teams $40/seat) but the unit changed — plans now sell agent-request limits against a first-party model pool instead of a dollar credit balance, and Cursor doubled those limits across all individual and Teams plans on July 21, 2026. Added Grok 4.5 (released July 8, 2026; trained jointly with SpaceXAI on Cursor data) to the versions list; Composer 2.5 remains as a separate, smaller weight class. Renamed Business to Teams Standard to match the pricing page. SpaceX acquisition still pending, expected Q3 close.
- Verified: individual plans unchanged. The June Teams update took effect July 1 for renewing customers — Business seats now split usage into two pools (first-party Composer/Auto vs third-party API models), giving more headroom at the same $40 price. Added Composer 2.5, Cursor's first-party model, to the versions list. Ownership note: SpaceX agreed June 16 to acquire Anysphere (Cursor) for $60B all-stock, expected to close ~Q3 2026 pending regulatory approval; reporting says Cursor usage data would feed Grok's training — added as a trust-boundary consideration.
- Verified plans: Hobby free, Pro $20, Pro+ $60, Ultra $200, Business $40/seat with $120 Premium seats (June 2026 update).
Give Cursor a job
The hands-on coding seat: an editor where AI agents write, refactor, and navigate the codebase with you.
Try Cursor →Alternatives with different jobs
GitHub Copilot
03 CreationThe completion seat: always-on code suggestions in your existing editor, plus agent work metered by AI credits.
Claude
02 DecisionsThe thinking seat: long documents, hard reasoning, code, and anything where the words have to be right the first time.
Replit
03 CreationThe build-and-ship seat: prototyping internal tools and small apps with an AI agent, hosted the moment they exist.