Cursor
The sharpest everyday coding editor, if you can live with the credit pool.
Best for: daily drivers who want a fast IDE with real model choice.
What we tested
Honest method. We did not create a paid account or run a controlled benchmark for this report. We reviewed the free tier, the product documentation, and the public pricing and feature pages ourselves, without signing up. Feature and pricing claims below are based on public documentation and pricing as of October 1, 2026, not on hands-on paid testing. Scores are editorial verdicts grounded in those documented capabilities, not measured benchmark results.
We will upgrade this report to full hands-on testing once the publication has paid subscriptions running.
Score breakdown
| Criterion | Points | Why |
|---|---|---|
| Code quality | 20 / 25 | Tab completion is consistently praised as best-in-class for in-editor work, and model choice lets you route hard tasks to stronger models. |
| Agentic coding | 17 / 20 | Agent mode plans and executes multi-step edits, and background cloud agents handle tasks async. Billing those runs against credits is the tax on the feature. |
| Codebase understanding | 12 / 15 | Good semantic search, but you still @mention files yourself. Windsurf's automatic indexing edges it here. |
| Speed and reliability | 8 / 10 | Fast and stable as an editor; heavy agent days can burn the credit pool, which is a reliability question in practice. |
| Value for money | 12 / 15 | Pro at $20 matches the market price, and unlimited Auto mode is generous. The credit pool adds cost anxiety that flat-rate plans avoid. |
| Privacy and data controls | 11 / 15 | Privacy mode is available to stop code from being stored, but the strongest controls (SSO, audit logs) sit on Teams and Enterprise. |
Strengths
- Best tab completion in the category for day-to-day editing flow.
- Real model choice: Claude, GPT, Gemini, and Cursor's own Composer models in one editor.
- Unlimited Auto mode on paid plans, so most users never touch the credit pool.
- Background agents that work a task in the cloud and return a PR.
- VS Code foundation means extensions, keybindings, and workflows transfer over.
Weaknesses
- The credit pool punishes manual model selection at API rates, with an ~8x cost swing between cheapest and priciest models.
- Billing has changed twice in two years, so long-term cost planning is guesswork.
- Context still needs manual @mentions rather than fully automatic codebase indexing.
- Teams pays double Pro per seat ($40) mostly for admin controls.
Pricing
| Plan | Price | Notes |
|---|---|---|
| Hobby | Free | Limited completions and agent requests. For evaluation only. |
| Pro | $20/mo | $20 credit pool, unlimited Auto mode, frontier models. |
| Pro+ | $60/mo | $60 credit pool, for heavy daily coders. |
| Ultra | $200/mo | $200 credit pool, for full-time agentic development. |
| Teams | $40/user/mo | SSO, admin controls, centralized billing. |
| Enterprise | Custom | Audit logs, pooled usage, account manager. |
Source: Cursor pricing page, checked Oct 1, 2026. Annual billing discounts roughly 20%. How billing works: since June 2025 each paid plan includes a monthly credit pool equal to its price; manually selecting a frontier model draws from the pool at that model's API rate.
Who should buy it
Individual developers who live in their editor and want the best inline completion plus the freedom to pick models per task. Also strong for anyone already comfortable with VS Code who wants AI depth without learning a new IDE.
Who should skip it
Teams with compliance needs should price out Teams or Enterprise rather than stacking Pro seats, and cost-anxious users who dislike usage meters will be happier with a flat-rate or quota plan. If most of your work is GitHub PR review and completions, Copilot costs half as much.
Affiliate disclosure
The Lab Report is supported by affiliate commissions. If you buy a tool after clicking a link on this site, we may earn a commission at no extra cost to you. Rankings and scores are never for sale: vendors cannot pay for better placement or better scores, and our verdicts are written before any commercial discussion. Commission income never influences what we test or what we recommend.
No affiliate program is active for Cursor yet. The link below goes to the official homepage.