Claude Code vs Cursor vs GitHub Copilot (2026): Which AI Coding Tool Should You Pay For?

June 2026 quietly reshuffled the price tags on all three of the most popular AI coding tools. GitHub Copilot dropped premium requests entirely and moved to usage-based AI Credits on June 1. Cursor split its included usage into two separate pools and added a Premium team seat, rolling out to renewals from July 1. Anthropic, meanwhile, left Claude Code alone — flat subscription tiers, plus an opt-in overage valve. If you compared these tools even three months ago, your notes are stale.

A word on method before the numbers. This is a documentation-based comparison: every price below comes from an official pricing page, changelog, or vendor announcement, re-checked on July 11, 2026, with the source linked. We have not run a hands-on benchmark for this piece and we won't pretend otherwise. Where community experience matters, we cite the threads.

The verdict, up front

• Solo developer on a budget: GitHub Copilot Pro ($10/mo) — still the cheapest paid entry point, and code completions no longer touch your credit balance. Just know the credits meter runs fast on agent-heavy work, and new sign-ups are only gradually reopening.
• Solo developer who codes with agents daily: Claude Code on the Pro plan ($17/mo annual, $20 monthly) — flat pricing, a real terminal-native agent, and a spend-capped overage option instead of a surprise bill.
• Power user / agent-first workflow: Claude Code Max ($100–$200/mo) or Cursor Ultra ($200/mo). Pick Max if the agent is your workflow. Pick Ultra if you live inside the Cursor editor and want the multi-model buffet.
• Small team (5–50): Cursor Teams ($40/seat, Premium seat $120) — the June 2026 split usage pools plus per-seat spend alerts make team cost governance genuinely easier than the alternatives.
• Enterprise: GitHub Copilot Business/Enterprise ($19/$39 per user) — pooled org-level credits, IP indemnity, and the GitHub-native admin surface still win at scale, provided you set budgets on day one.

The rest of this article shows the working behind those calls.

What actually changed in June–July 2026

Three separate repricing events landed within about five weeks of each other. That's why so much older comparison content is now wrong.

GitHub Copilot: premium requests are dead, AI Credits are in. On June 1, 2026, every Copilot plan moved to usage-based billing. Premium request units (PRUs) were replaced with GitHub AI Credits, consumed by actual token usage — input, output, and cached tokens — at published per-model rates (GitHub Blog announcement, June 1, 2026; changelog). Monthly subscribers migrated automatically; annual subscribers keep PRU pricing until their term expires. Two less-advertised changes matter here. The automatic fallback to cheaper models was discontinued. And Copilot code review now consumes GitHub Actions minutes on top of AI Credits. The launch was rocky — as of July 11, new sign-ups for Pro, Pro+, and Max are still only gradually re-enabling per the plans page, and the community discussion filled with reports of allowances draining far faster than under PRUs. Notably, GitHub has already adjusted upward: the June 1 announcement listed $10 of included credits on Pro, but the plans page as of July 11 shows $15 included plus a $5 "flex" allotment.

Cursor: two usage pools and a Premium team seat. In June 2026, Cursor announced that every Teams seat now carries two separate pools of included usage — one for first-party models (Auto and Composer 2.5) and one for third-party API models (Claude, GPT, Gemini and friends) — effectively more headroom at the same seat price. A new Premium seat launched at 5x the Standard seat's included usage for 3x the cost. New customers got the change immediately; renewing customers move over on billing cycles starting July 1, 2026 (Cursor blog, June 2026). The same split-pool logic now shows up in the individual plan docs too, where first-party models (Auto, Composer 2.5, Grok 4.5) get "significantly more included usage" than the dollar-denominated API pool (Cursor docs, checked July 11, 2026).

Claude Code: steady tiers, opt-in overage. Anthropic didn't reprice this summer — which is itself a differentiator. Claude Code is included from the Pro plan up: Pro runs $17/month billed annually ($200 up front) or $20 monthly, and Max starts at $100/month with a choice of 5x or 20x Pro usage (claude.com/pricing, checked July 11, 2026). Hit your plan's limit and you can enable "usage credits" — a prepaid balance spent at standard API rates with an adjustable monthly spend limit — rather than being upsold a tier (Claude Help Center, checked July 11, 2026).

Pricing at a glance (as of July 11, 2026)

All prices USD, monthly billing unless noted. "Included usage" reflects what each vendor's official page states today.

What each tool is actually selling

GitHub Copilot is the incumbent inside VS Code, JetBrains, and github.com itself. The thing it now sells cheapest — unlimited code completions and next-edit suggestions on any paid plan, exempt from credit consumption — is genuinely unlimited. Everything agentic draws down AI Credits at token-based rates: chat with premium models, coding agent runs, code review. Fair deal for light users. Volatile for heavy ones — community reports after June 1 describe monthly allowances evaporating in days on agent workloads (community discussion). GitHub's answer is budget controls at the user, cost-center, and enterprise level, plus the new $100 Max tier for "high-volume agent workflows."

Cursor sells an editor, and the editor is the point. You get frontier models from multiple vendors in one place, an agent (Composer) that Cursor subsidizes through its first-party pool, and — since June — a much clearer answer to "where did my usage go?" The dual-pool design means routine agent work on Auto or Composer 2.5 barely dents your budget, while pointing the agent at Claude or GPT burns the dollar-denominated API pool at provider rates. Credit where due: Cursor's own docs offer honest sizing guidance. Tab-completion-centric users fit in the $20 plan, daily agent users typically land at $60–$100 a month all-in, and automation-heavy power users exceed $200 (Cursor docs).

Claude Code is a terminal-native agent rather than an editor — you bring your own IDE, or none. Its pricing pitch is predictability: a flat plan sized by usage multiplier, with overage strictly opt-in through prepaid usage credits under a spend limit you set (Claude Help Center). The trade-off is single-vendor. You're buying Anthropic models only, and if your team wants model diversity, this isn't where you get it. For teams, Anthropic's pricing page distinguishes standard seats ($20–25) from premium seats ($100–125) — the latter carrying the heavier usage allowance that agent-driven development realistically needs.

Decision matrix by developer profile

Two things cut across every row of that table. These tools aren't mutually exclusive — a Copilot Business org can and often does run Claude Code or Cursor alongside it, since Copilot's unlimited completions cost nothing extra to keep. And whichever you pick, the June 2026 shift means the sticker price is no longer the real price for heavy users. The included-usage column above is the number to compare.

The honest cost math

Because Copilot credits are token-denominated at per-model rates, and Cursor's API pool bills at provider rates, a "month of AI coding" is now a consumption question, not a subscription question. A rough documentation-derived frame:

• Light use (completions, occasional chat): Copilot Pro's $10 is the floor. Its unlimited completions are the best pure value in the market right now.
• Moderate agent use (a few agent sessions daily): Claude Code Pro or Cursor Pro will both hit limits some days. Claude's opt-in credits and Cursor's PAYG both work; Claude's spend cap is the more conservative default.
• Heavy agent use (agents running most of the working day): you are realistically in $100–200+ territory on any of the three — Claude Max, Cursor Ultra, or Copilot Max. At that spend, tool fit should decide, not price. Terminal agent? Claude. Multi-model editor? Cursor. GitHub-integrated workflow? Copilot.

One asymmetry the vendors won't flag for you: Copilot's discontinued model fallback means there's no longer an automatic cheaper path when credits run low — you either buy more or stop. Claude and Cursor both degrade more gracefully (opt-in credits with a cap; PAYG at unchanged rates).

FAQ

Did GitHub Copilot get more expensive in June 2026? Base plan prices didn't change — Pro is still $10, Business $19, Enterprise $39. What changed is how usage is metered: premium requests became token-based AI Credits, and heavy agent users report burning through allowances much faster than before (GitHub Blog; community thread). For completion-focused users, effectively nothing changed.

Is Claude Code included in the free Claude plan? No — as of July 11, 2026, Anthropic's pricing page lists Claude Code starting on the Pro plan, not the Free tier (claude.com/pricing). If you want to evaluate the Claude Code workflow, the $20 monthly Pro plan (no annual commitment) is the realistic entry point.

What are Cursor's "usage pools" and why do they matter? Since June 2026, Cursor tracks first-party model usage (Auto, Composer 2.5) separately from third-party API usage (Claude, GPT, Gemini). The first-party pool is much larger, so keeping routine agent work on Cursor's own models stretches a plan dramatically, while third-party models bill against a dollar pool at provider rates (Cursor blog; docs).

Can I still buy an annual Copilot plan with the old premium-request pricing? Existing annual Pro and Pro+ subscribers keep premium-request pricing until their plan expires (changelog). For new customers it's a different story: sign-ups for Pro, Pro+, and Max are only being "gradually" re-enabled per the plans page as of July 11, 2026 (plans page).

Which is cheapest for a 10-person team? On sticker price: Copilot Business at $19/user ($190/mo, plus a $30/user credit promo through August 2026). On realistic agent-heavy usage: Cursor Teams often wins, because Standard seats ($40) cover most members while only heavy users need Premium seats ($120), and admins get threshold alerts before overages land. Claude Team's mixed seating ($20–25 standard, $100–125 premium) follows the same logic with Anthropic-only models.

Do any of these prices include the underlying model API costs? Copilot credits and Cursor's API pool both meter third-party model usage at published rates within your plan's included amount — you're prepaying API consumption, effectively. Claude plans include usage in the subscription; only the opt-in usage credits bill at API rates (Claude Help Center).

Keep exploring on ToolPick

• Cursor alternatives worth a look
• GitHub Copilot pricing, tracked over time
• Claude Code review: what it does well and where it stalls (deep-dive coming soon)
• Cursor vs Windsurf comparison
• AI tool cost calculator — model your monthly spend
• This week's AI tool updates

Methodology and disclosure

This article is a documentation analysis, not a hands-on benchmark. All prices and plan mechanics were taken from official vendor pricing pages, changelogs, help-center articles, and vendor blog posts, each re-fetched on July 11, 2026, plus one public community discussion for post-launch user reports. Five representative claims were independently re-verified against their cited sources during editorial review on the same date; two were corrected as a result. Vendors change pricing without notice — if a number above disagrees with a vendor page you're looking at, trust the vendor page and check our tracked pricing pages for the update. No vendor sponsored or reviewed this comparison.

Sources

1. GitHub — "GitHub Copilot is moving to usage-based billing," June 1, 2026. https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/ (accessed 2026-07-11)
2. GitHub Changelog — "Updates to GitHub Copilot billing and plans," June 1, 2026. https://github.blog/changelog/2026-06-01-updates-to-github-copilot-billing-and-plans/ (accessed 2026-07-11)
3. GitHub — Copilot plans and pricing page. https://github.com/features/copilot/plans (accessed 2026-07-11)
4. GitHub Community — discussion #192948 on usage-based billing. https://github.com/orgs/community/discussions/192948 (accessed 2026-07-11)
5. Cursor — "Improvements to Teams Pricing," June 2026. https://cursor.com/blog/teams-pricing-june-2026 (accessed 2026-07-11)
6. Cursor — Models & Pricing documentation. https://cursor.com/docs/models-and-pricing (accessed 2026-07-11)
7. Cursor — Pricing page. https://cursor.com/pricing (accessed 2026-07-11)
8. Anthropic — Claude plans and pricing. https://claude.com/pricing (accessed 2026-07-11)
9. Anthropic — "Manage extra usage for paid Claude plans," Claude Help Center. https://support.claude.com/en/articles/12429409-manage-extra-usage-for-paid-claude-plans (accessed 2026-07-11)