AI tools
GitHub Copilot's new pricing, explained — will you pay more?
The popular AI coding assistant changed how it charges. Here's what it means for your bill.
The answer
From 1 June 2026 Copilot charges by usage, not a flat monthly fee.
If you or your team use GitHub Copilot to help write code, its pricing just changed — and you might be wondering whether your bill is going to get bigger. The honest answer: it depends on how you use it. Here's the plain-English breakdown, without the jargon.
What actually changed, step by step
Before 1 June 2026, you paid one flat monthly fee and could use Copilot as much as you liked — a bit like an all-you-can-eat buffet. From that date, GitHub switched to what it calls 'flex' billing. The monthly fee is still there, but now it's the starting point, not the limit. If you use Copilot lightly — quick code suggestions, occasional questions — you probably won't notice much. But if you're using its more powerful features, especially the ones that work automatically across your whole codebase ('agent' features), the amount you use is now tracked and billed on top.
GitHub also added a new $100-a-month 'Max' plan for people who expect to use Copilot heavily. Think of it as a bigger buffet allowance, but at a higher upfront cost. Here's a simple comparison:
| Before 1 June 2026 | From 1 June 2026 | |
|---|---|---|
| Monthly fee | Fixed — covers everything | Entry point — usage on top |
| Light use | Same price regardless | Similar to before |
| Heavy/agent use | Same price regardless | Costs more |
| Max plan ($100/mo) | Didn't exist | New option for heavy users |
| Bill predictability | Completely predictable | Varies with your usage |
Who will actually pay more?
If you mainly use Copilot for simple things — it autocompletes your code as you type, or you ask it quick questions — you'll probably end up paying similar to what you did before. The extra charges kick in when you use the heavier features: things like asking Copilot to go off and independently review your whole codebase, write a batch of tests, or fix several files at once without you guiding every step. Those tasks use a lot more computing power, and that's what the new billing reflects. A helpful rule of thumb: the more you let Copilot run on its own without you typing, the more it costs under the new system.
GitHub moved Copilot to a usage-based model from 1 June 2026, with the plan price now acting as an entry point and a new $100 Max tier added for higher consumption — a shift that drew immediate developer backlash over less predictable bills.
Why is this happening?
Running AI costs money every single time you use it. When you ask Copilot to suggest the next word in your code, that's a small request. When you ask it to autonomously review and improve fifty files, that's a lot of computer time. The old flat-fee model didn't distinguish between those — everyone paid the same, which meant that low-usage people were quietly subsidising high-usage ones. GitHub's new model tries to charge you more in line with what you actually use.
And it's not just GitHub. Windsurf (another AI coding tool) rebranded to Devin Desktop and changed its pricing around the same time. Google's Gemini CLI (a coding assistant from Google) moved to a new brand called Antigravity, again with pricing changes. This is a shift happening across the whole industry in 2026 — not a single company being awkward, but an entire category arriving at the same conclusion: the AI tools that were once nearly free to use at any level are starting to charge closer to their real costs.
The June 2026 AI coding landscape shows the entire market moving away from flat monthly fees: Claude Code, Antigravity (formerly Windsurf), Codex and Copilot are all repricing or repositioning, with agentic-intensity as the key cost driver.
What you should do next
You don't need to rush into any decision. For most developers using Copilot casually, the change will be small. But it's worth spending five minutes checking your Copilot usage stats — GitHub usually shows these in your account settings — so you know what you're working with. If you're a heavy user who relies on the agent features, have a look at the $100 Max plan to see if the predictability of a higher flat rate is worth it for you. And if you'd like to compare other AI coding tools before committing, options like Cursor and newer entrants are also worth a look — the market is genuinely competitive right now, and there's no harm in comparing.
Frequently asked questions
Should I cancel Copilot?
What is the new $100 Max plan?
Are other AI coding tools doing the same thing?
What features cost the most under the new billing?
How do I check my current Copilot usage?
Sources
- AI Coding Tools Pricing: The June 2026 Reality Check — Developers Digest, 1 June 2026
- AI Coding Agents 2026: Claude Code vs Antigravity vs Codex vs Cursor vs Copilot — Lushbinary, 5 June 2026
- GitHub Copilot pricing and plans — GitHub, 1 June 2026