The cheap AI plan was always too good to last. Now the bill is catching up.
A growing list of tech firms are dropping flat monthly fees. They charge by usage instead, so the more you ask the AI to do, the more you pay.
What's Changing
GitHub, owned by Microsoft, is moving its Copilot coding tool to usage-based billing in June 2026. Heavy work now means a bigger bill.
The thing they charge for is the token. A token is a small chunk of text the AI reads or writes.
A quick question costs little. A long job can run into hundreds of thousands of tokens.
The price tags can sting. Some developers say their monthly bill jumped from $29 to $750 after the switch, and a few report even larger jumps.
GitHub also paused new signups for some plans in April. It blamed weekly computing costs that had nearly doubled since January.
We explain shifts like this and what they mean for your money in Market Briefs, and you get a free investing masterclass when you sign up.
Anthropic Felt The Same Squeeze
GitHub isn't alone. Anthropic briefly pulled its Claude Code tool from the $20 plan in April, then put it back after users pushed back.
The firm kept testing the change on a few new signups. Even popular tools are straining under flat pricing.
OpenAI made a similar move around the same time. It raised the price it charges coders to use its newest model.
The pattern is clear across the field. Cheap, all-you-can-use AI is getting harder to find.
Why The Flat Fee Broke
A flat fee works like a gym membership. The math only holds up if most people don't show up every day.
The heaviest AI users now show up all the time. For them, the real cost can run up to 10 times what their fee covers.
Picture a power user on a $200 plan. That person can burn through $600 to $1,500 in tokens in a single month.
That turns every power user into a money-loser. Spread across millions of people, those losses pile up fast.
What To Watch
This is bigger than coding tools. The same pay-per-use shift is spreading to AI that writes emails and builds slide decks.
AI is also writing more of the work itself. Even inside big tech firms, machines now write a growing share of new code.
For investors, it's a signal worth noting. The days of AI firms buying growth with cheap prices are fading.
How these firms price their tools will shape how much they earn. Some users will cut back as bills rise, while others pay up for the speed.
The new test is plain: can these products turn a profit? The free trial for the whole industry is ending.
Want to follow the money behind the AI story? Get Market Briefs every weekday morning and you'll also pick up a free 45-minute investing course as a bonus.
