Rewritten 20 August 2026. Every price below was checked against the vendor's own pricing page on that date — not recalled, not estimated. The full deep-dive version of this comparison is being re-verified number-by-number on stackwrite.com and will return; until then, this shorter version only states what I could confirm today.
The three tools solve different problems, and the pricing pages have all changed this year. Here is where they actually stand.
Verified pricing (20 August 2026)
| Tool | Free option | Individual paid | Top tier |
|---|---|---|---|
| Claude Code | No free tier — needs a subscription or API key | Pro $20/mo | Max from $100/mo |
| Cursor | Hobby (limited agent requests) | Pro $20/mo | Ultra tiers above Pro; Teams $40/user/mo |
| GitHub Copilot | Free plan ($0, limited chat/agent use) | Pro $10/mo | Pro+ $39/mo, Max $100/mo |
Sources: claude.com/pricing, cursor.com/pricing, github.com/features/copilot/plans — each checked the day this was written.
Two changes worth knowing if you last looked months ago:
- Copilot added Pro+ ($39) and Max ($100) tiers, priced as bundled monthly model credits ($70 and $200 worth respectively), and its agent mode can now hand work to third-party agents including Claude Code and Codex. The $10 Pro plan is unchanged.
- Claude Code on a subscription is metered by usage allotment, and the Max plan starts at $100/mo — I have seen it quoted as $20 in several roundups (including, previously, my own), which is the Pro price, not Max.
Which one, honestly
Claude Code if your work is multi-file: it reads the repo, edits across files, runs tests. It lives in the terminal, which is either its best feature or a dealbreaker for you — there is no middle.
Cursor if you want the strongest in-editor experience. It's a VS Code fork, so the switching cost is near zero, and the tab-completion model is still the best of the three for staying in flow.
Copilot if you want the lowest price that covers the basics, or you're already deep in the GitHub ecosystem. The free plan is a real (if tight) way to evaluate agent-style coding at $0.
If you're choosing for a team: Cursor's per-seat price doubles its individual price, Copilot's doesn't. Run the seat math before the feature comparison.
Disclosure: I run stackwrite.com, where the fully-sourced version of this comparison is being re-verified. I use all three tools; none of these links are affiliate links.
Top comments (0)