I’ve subscribed to GitHub Copilot since the early access days. Over time my workflow came to depend on it: always there, always fast, always finishing my sentences. I renewed without thinking about it.
Then one morning in late April I hit a wall mid-session. It wasn’t a free-tier rate limit or an outage. It was a hard weekly usage cap, on a paid Pro subscription.
What changed
On April 20, 2026, GitHub published a blog post titled “Changes to GitHub Copilot Individual plans”. The headline items:
- New sign-ups for Pro, Pro+, and Student plans were paused
- Weekly token-based usage limits were tightened for all individual plans
- Session limits were introduced on top of the weekly ones
- Opus models were removed from Pro plans entirely
The justification was reasonable on its face. Agentic workflows have gotten expensive, long-running sessions are burning through infrastructure, and a handful of heavy users were making the service unreliable for everyone else. Those are real problems.
What stings is that I didn’t sign up for a product with session limits and weekly token caps. They weren’t part of the deal when I subscribed. They were added to a plan I was already paying for, without notice, and explained afterwards in a blog post most users wouldn’t see.
The current pricing page has this in the footnotes:
Response times may vary during periods of high usage. Requests may be subject to rate limiting.
That footnote is covering for a very different product from the one that was sold.
The usage-based billing pivot
A week later GitHub dropped another shoe: Copilot is moving to usage-based billing on June 1, 2026. Premium request units are being replaced by “GitHub AI Credits,” consumed by token usage (input, output, and cached tokens) at published API rates per model.
Pro stays at $10/month, but that $10 now buys $10 of AI Credits instead of an effectively unlimited flat-rate subscription. If you’re using Claude Sonnet or GPT-5 for a long agentic session, the credit disappears fast. The price stayed the same while the product changed a lot.
I understand why they’re doing it. A multi-hour autonomous coding session obviously costs more than a quick chat question, and charging one flat rate for both was never sustainable once agentic usage took off. GitHub absorbed that cost for a while and now can’t.
But changing pricing for new customers is different from changing what existing subscribers get. Annual plan holders are moved off their plans at expiry, with no grandfathering of the model they signed up for. Monthly subscribers are migrated automatically on June 1 whether they opt in or not.
Why the limits felt arbitrary
My specific frustration is that the new limits are opaque and inconsistent.
Two separate constraint systems run in parallel: session limits (per-request windows, to protect infrastructure) and weekly limits (a total token budget, for cost control). They’re completely independent, so you can have premium requests left and still hit a usage limit. The dashboard can show requests remaining while you’re blocked from the model you want.
GitHub now shows a warning in VS Code when you reach 75% of your weekly limit, which is an improvement. But getting told you’re low on a budget you didn’t know existed, on a subscription you bought before that budget was introduced, feels like the product getting worse and calling it transparency.
What I moved to
I switched to Claude Code. It’s a terminal-first tool built on Anthropic’s Claude models and designed for agentic development sessions, the long, multi-step, repo-spanning work that was breaking Copilot’s pricing model.
The billing is transparent upfront. You pay for what you use at published, consistent rates. There’s no dual-limit system to reason about, no premium request concept separate from a token budget, and no footnotes saying what you pay for might be rate limited.
It’s also very good at the agentic work: reasoning through multi-file changes, writing tests, and refactoring across a codebase, beyond completing inline suggestions. That’s what I was pushing Copilot to do, and it’s what Claude Code is built for.
The broader point
I’m not trying to pile on GitHub. Copilot is impressive engineering, and the team is dealing with a real cost problem caused by how fast agentic usage has grown. Usage-based billing is probably the right call long term.
But when a product changes its core terms mid-subscription (what you get, which limits apply, which models you can use) without telling subscribers beforehand, trust erodes. I found out about my weekly token limit by hitting it, not from a changelog.
The developer tools market is competitive enough now that trust matters. If you tell me I’m on a premium plan, I should be able to rely on what that means for the period I paid for. When I can’t, I start looking at alternatives, and right now the alternatives are pretty good.