OpenAI will halve the usage allowance on its 200$ plan from October 30 and has launched a new 500$ plan. According to The New Stack, the plan that costs 200$ will give you half of what it did before.
What changed
The standard plan costs 20$. Think of it as 1x. The 200$ plan used to give 20x that. From October 30 it gives 10x. Same price, half the usage.
OpenAI also launched a 500$ plan with 25x the usage and an Ultrafast mode for GPT-6 Astra that runs up to 8x faster in Codex. So for 500$ you get 25% more usage than the old 200$ plan, but you pay 2.5 times more.
OpenAI's explanation is that cheaper models mean you get more work done anyway.
We have seen this move before
Apple took the charger out of the iPhone box and said it was for the environment. You still needed a charger. You just paid for it separately.
This looks like the same move. Remove the thing, call it progress, sell it back as an upgrade.
Why I expect prices to keep rising
These models cost real money to run, and a 20$ flat plan cannot carry that forever. The era of cheap, flat-rate AI plans looks like it is ending.
The fair question for every founder is whether other providers, Anthropic included, will follow. You do not need the answer today. You need a budget that survives either one.
What founders should do now
- List every AI subscription your team pays for, and what each one is used for.
- Put a price rise into your budget today, even a rough one.
- Check which tasks really need the top model and which can run on a cheaper one.
- Do not build a process that only works on one tool at one price.
The same logic applies to AI agents: set limits before you scale, as in AI agents need a leash.
Different market, same door. You still have to knock right on LinkedIn.