Key Takeaways
- Entry pricing: GitHub Copilot Pro is $10/month, Cursor Pro is $20/month, and Claude Pro (which includes Claude Code) is $20/month, per each company's official pricing page.
- Copilot is the only one that states code completions are unlimited — GitHub's docs say completions and next edit suggestions "are not billed in AI credits" on all paid plans.
- Top individual tiers land at the same number by coincidence: Cursor Ultra is $200/month and Claude Max 20x is $200/month, while Copilot Max tops out at $100/month.
- Cursor and Copilot both meter third-party models at API rates — both list Claude Opus 5 at $5.00 per million input tokens and $25.00 per million output tokens.
The cheapest of the three is GitHub Copilot at $10 a month, and it is the only one of the three that promises unlimited code completions on every paid plan. But if you want an agent that runs commands and rewrites files all day, the entry tier that matters is $20 — and Cursor, Claude Code, and Copilot each spend that $20 in a completely different way. Cursor meters you at raw model API prices, Claude Code gives you a session-based allowance shared with the chat app, and Copilot hands you a fixed pile of credits where 1 credit equals exactly one cent.
The one-table version
Every figure below comes from the vendor’s own pricing page or docs, checked in August 2026.
| Tool | Free tier | Entry paid | Mid tier | Top individual | Team / org |
|---|---|---|---|---|---|
| Cursor | Hobby — “Limited Agent requests,” access to Composer | Pro $20/mo — “Extended limits on Agent” | Pro Plus $60/mo — “3x Pro limits on Agent” | Ultra $200/mo — “20x Pro limits on Agent” | Teams $40/user/mo (Standard), $120/user/mo (Premium, “5x Standard limits”); Enterprise custom |
| Claude Code | Free plan is chat only — no Claude Code | Pro $20/mo ($17/mo billed annually at $200 up front) | Max 5x $100/mo — “five times more usage per session than the Pro plan” | Max 20x $200/mo — “20 times more usage per session” | Team seats $20/mo annual or $25/mo monthly (2–150 people); Premium seat $100/mo annual, “5x more usage than standard seats”; Enterprise $20/seat plus usage at API rates |
| GitHub Copilot | Free $0 — “2000 completions and 50 chat requests (including Copilot Edits)” | Pro $10/mo — 1,000 base + 500 flex = 1,500 AI credits | Pro+ $39/mo — 3,900 base + 3,100 flex = 7,000 credits | Max $100/mo — 10,000 base + 10,000 flex = 20,000 credits | Business $19/seat/mo (1,900 credits); Enterprise $39/seat/mo (3,900 credits) |
One caveat on that last row: GitHub’s docs note that self-serve purchases of Business and Enterprise plans were “temporarily paused” on April 22, 2026, with sign-ups expected to reopen. If you are buying for a team today, check availability before you plan around those seat prices.
Cursor: an editor that bills at API cost
Cursor’s pricing docs describe two separate monthly usage pools. The Cursor Models pool covers its own and partner models — Grok 4.6, Grok 4.5, and Composer 2.5 — with what the docs call “significantly more included usage.” The other models pool covers third-party frontier models from OpenAI, Anthropic, and Google, and those are “charged at the model’s API price.”
That second sentence is the whole Cursor pricing model in eight words. You are not buying requests; you are buying a prepaid balance of tokens metered at published per-million-token rates. Cursor’s docs price Claude Opus 5 at $5 input and $25 output per million tokens — the same rate GitHub publishes for the same model.
What happens when you burn through it is unusually clear for this category. Cursor’s docs say you can “Add on-demand usage” and “Continue at the same API rates with pay-as-you-go billing,” and — pointedly — that “Requests are never downgraded in quality or speed.” No silent throttling, no mystery slow mode. You either stop or you keep paying the meter.
The gap in Cursor’s disclosure is the number itself. The tiers are described relative to each other — Pro Plus is “3x Pro limits on Agent,” Ultra is “20x Pro limits” — but the docs do not publish how many dollars or tokens of included usage sit inside Pro. You know the multiplier; you do not know the base. That is worth knowing before you budget, and it is a fair chunk of why Cursor cost estimates vary so wildly between developers. We go deeper on the day-to-day experience in our Cursor review.
Also on the pricing page: MCPs, skills, and hooks, cloud agents, and Bugbot on usage-based billing all sit on Pro and above. Teams adds SAML/OIDC SSO, privacy mode, usage analytics, and centralized billing; Enterprise adds pooled usage, SCIM, audit logs, and API access.
Claude Code: one subscription, many surfaces
Claude Code is the odd one out because it is not really a separate product line — it is a feature of a Claude subscription. Anthropic’s pricing page lists “Includes Claude Code” as a Pro benefit, and the Team plan description says it “Includes Claude Code and Claude Cowork.” The Free plan does not include it at all.
That bundling cuts both ways. On the plus side, one $20 Pro subscription covers the chat app, the desktop app, and the coding agent. Anthropic’s docs describe Claude Code as an agentic tool that “reads your codebase, edits files, runs commands, and integrates with your development tools,” available “in your terminal, IDE, desktop app, and browser,” with plugins for VS Code and JetBrains plus GitHub Actions and GitLab CI/CD integrations. There is even a Cursor install target for the extension, which tells you something about how these products actually coexist.
On the minus side, the budget is shared. Anthropic’s support docs are blunt about it: Pro and Max plans “offer usage limits that are shared across Claude and Claude Code, meaning all activity in both tools counts against the same usage limits.” A long research chat in the morning eats into your afternoon coding session.
The limit structure is session-based. Anthropic’s Max article says limits reset every five hours, with a weekly usage cap across all models on top. Max 5x at $100/month gives “five times more usage per session than the Pro plan,” and Max 20x at $200/month gives “20 times more usage per session.” Anthropic does not publish a token or message count for any of these tiers, so the honest summary is that you learn your own ceiling by hitting it.
GitHub Copilot: unlimited autocomplete, metered agents
Copilot is the only one of the three with a genuinely useful free tier for daily work — 2,000 completions and 50 chat requests a month, with agent mode and the CLI included, though code review is not.
Its paid model splits cleanly in two. Completions are free forever: GitHub’s docs state that “Code completions and next edit suggestions are not billed in AI credits. They remain unlimited for all paid Copilot plans.” Everything agentic — Copilot Chat, Copilot CLI, the cloud agent, Copilot Spaces, Spark, and third-party coding agents — draws on AI credits, where “1 AI credit = $0.01 USD.”
That exchange rate makes Copilot the easiest of the three to model. Pro’s 1,500 credits are $15 of model usage on a $10 subscription. Pro+ at $39 gets you $70 of usage. Max at $100 gets $200. In each case you are buying model tokens at a discount to what they cost, priced per million tokens by model — GPT-5 mini at $0.25 input and $2.00 output, Claude Opus 5 at $5.00 and $25.00.
Pro+ adds one feature worth flagging for anyone comparing these three: the ability to “Delegate tasks to third-party coding agents like Claude by Anthropic and OpenAI Codex (Preview).” Copilot is quietly positioning itself as the place you dispatch the other two from.
So which one do you actually pay for?
If you write code in bursts and mostly want fast autocomplete, Copilot Pro at $10 is hard to argue with — unlimited completions, and the credits are gravy. If you live inside an agent that runs commands and edits files across a repo all day, the $20 tiers are the real comparison, and the deciding factor is your tolerance for a variable bill: Cursor’s meter keeps running at API cost, while Claude Code’s cuts you off until the window resets. Neither is strictly better. One risks a surprise invoice, the other risks a surprise wall at 4pm.
If you are still shortlisting, our roundup of the best AI coding assistants of 2026 covers the tools that did not make this three-way, and our look at browser agents and automated code review covers the workflow layer sitting on top of all of them.
What happens next
The convergence in these pricing pages is the story. Two of the three now meter third-party models at published per-token API rates, at identical prices for the same model — which means the subscription fee is increasingly paying for the harness, not the intelligence. Expect the differentiation to keep moving toward what the tool does with a model rather than which models it can reach. Copilot Pro+ already lets you dispatch work to Claude and Codex; Anthropic already ships a Cursor extension. The tidy “pick one” framing is eroding, and the plausible 2027 setup is two subscriptions doing different jobs rather than one winner.
Quick poll
Which pricing model would you rather live with?
Cursor's docs say on-demand usage continues "at the same API rates" and that "Requests are never downgraded in quality or speed"; Anthropic's Max limits instead reset on a five-hour session window.
FAQ
Is Claude Code included in a $20 Claude Pro subscription? Yes. Anthropic’s pricing page lists “Includes Claude Code” under Pro at $20/month, or $17/month if you pay $200 annually up front. The Free plan does not include it.
Which is cheapest for pure autocomplete? GitHub Copilot. Its Free plan includes 2,000 completions a month, and GitHub’s docs state completions and next edit suggestions are unlimited and not billed in AI credits on every paid plan, starting at $10/month.
Can I run out of usage mid-task? On Cursor, no — the docs say you can add on-demand usage at the same API rates and requests are never downgraded. On Claude Pro and Max you can, because limits reset on a five-hour session window with a weekly cap on top, and chat usage counts against the same pool.
What does a GitHub AI credit actually cost? Exactly one cent. GitHub’s billing docs state “1 AI credit = $0.01 USD,” so Copilot Pro’s 1,500 monthly credits represent $15 of model usage on a $10 subscription.
