GitHub Copilot's June Billing Overhaul Is Real: Here's What Changes and Who It Hurts
GitHub paused new sign-ups for Copilot Pro, Pro+, and Student plans this week. That's the tell. Companies don't freeze new customer acquisition for a product that's hitting its numbers — they do it when demand is outrunning their cost recovery targets faster than they can reprice. The billing overhaul arriving June 1 isn't a feature update. It's an accounting correction, and everyone on a paid Copilot plan needs to understand what actually changes.
The short version: starting June 1, GitHub Copilot moves from a request-based model (buy a bundle of Premium Request Units, use them) to an AI Credits system (buy credits, spend them per interaction). The credits are consumed by Copilot Chat, Copilot CLI, Copilot cloud agent, Copilot Spaces, Spark, and third-party coding agents. Code completions and next-edit suggestions remain unlimited on all paid plans. The rate-limit dashboard in VS Code and Copilot CLI will show real-time usage against your credit allocation.
The longer version requires doing some math, because the model-specific pricing means your cost per interaction will vary significantly depending on which model you're using.
The Opus removal is the concrete pain point
As of this week, Opus 4.7 is only available on Copilot Pro+. Opus 4.5 and 4.6 have been removed from Copilot entirely. Pro subscribers who were using Opus for complex reasoning tasks — architecture planning, security-sensitive code review, intricate debugging — are now on a different model tier or paying $20 more per month for Pro+. GitHub's stated reason is that keeping Opus available on Pro would require rate limits that "result in a worse overall experience." That's honest, but it doesn't change the outcome: Pro users who needed Opus now need to find an alternative.
The community reaction in GitHub's discussion forums reflects genuine frustration from developers who subscribed specifically for Opus access. Several posts note that the $19/month Pro tier was worth it for Opus-powered reasoning tasks, and the upgrade to $39/month Pro+ changes the value calculation against alternatives like Cursor or direct API subscriptions. This isn't universal — developers using Copilot primarily for autocomplete are unaffected — but the chat-heavy power users are the ones generating the most discussion.
GitHub's billing preview tool, arriving in early May for existing subscribers, will show estimated costs based on April usage patterns. That's useful signal, but it's backwards-looking. If your April usage doesn't reflect your actual usage under the new model — because you were unconsciously rationing under PRU limits, or because your workflow shifted after a project ended — the preview may not accurately predict your June bill. Watch the usage dashboard starting in May and treat the preview as a range, not a guarantee.
Why the sign-up pause matters more than the billing change
Here's the thing about the sign-up pause: it's the most honest signal in the announcement. GitHub isn't saying "we're upgrading our infrastructure." It's not saying "we're adding features." It's saying demand has exceeded capacity or cost recovery targets, and they've decided to manage that constraint by limiting new paid seats rather than by raising prices immediately. For existing Pro and Pro+ subscribers, this might actually mean more capacity headroom in the near term. But it also means GitHub is managing a product that isn't generating the margin it expected at current price points.
The comparison to alternatives is worth doing explicitly. Cursor, Claude Code, and other AI coding tools have different billing models — some flat-rate, some usage-based, some API-key based. If you were on Copilot Pro primarily for the chat features and you're finding that the AI Credits model ends up costing more than a $20/month flat-rate alternative, the June 1 transition is a natural switching point. The refund window for existing Pro/Pro+ subscribers is open until May 20 for anyone who wants to exit before the new billing model takes effect.
What to actually do between now and June 1
If you're on a Copilot paid plan, three things are worth doing this week. First, check the billing preview when it drops in early May against your actual April usage — and then run a mental model of what your usage might look like under the new credits system, not just extrapolating linearly. If you were hitting PRU limits before, your chat usage will be higher under credits than your April data suggests. Second, evaluate whether your actual workflow is completion-heavy or chat-heavy. If you're primarily using Copilot for autocomplete, the billing change doesn't affect you much. If you're running daily chat sessions for code review or architecture discussions, watch the dashboard from May onward. Third, if you're on Pro and relying on Opus, the math on upgrading to Pro+ needs to be run explicitly against alternatives — including whether a direct API subscription or a different tool makes more sense for your use case.
The billing change is a re-set, not a catastrophe. But it's the first time Copilot has materially changed the economic proposition for paying subscribers, and the sign-up pause tells you this is structural, not temporary. The "unlimited AI coding assistant for $10-19/month" era is over. What replaces it depends on whether the usage-based model ends up costing more or less than the competition — and that's going to be different for different usage patterns.
The developers who'll feel this most are the ones who adopted Copilot for complex reasoning work, not autocomplete. For everyone else, it's a watching brief — but one worth watching closely starting in May, before the June 1 switch flips.
Sources: GitHub Copilot Billing Documentation, GitHub Copilot Plans & Pricing, GitHub Community Discussion #192963