OpenAI halves what the $200 Pro plan buys
Tibo says the reopened plan is worth half the API dollars of the old one, and dax argues the people who should be angry are early stage startups.
OpenAI is reopening its $200 Pro subscription to new subscribers and changing how usage on it is calculated. Tibo posted the numbers ahead of the announcements, and they are not subtle. "if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan", he wrote.
if you do the math, it will net out at half the dollar in API spend compared to the old Pro $200 plan
The post is a rare case of a lab explaining a downgrade before shipping it. Tibo lists four reasons. There is a commitment not to reintroduce the 5 hour limit, so the weekly usage can be spent whenever you want. There is a promise that over time you get more work done at increasing quality, as models get more efficient and those gains get passed down as API price cuts. There is an explicit refusal to inflate API list prices so the subscription looks generous by comparison, with GPT-6 Sol and GPT-6 Luna introduced this week at 50% of their previous price as the evidence. And there is something unnamed being added to the subscription that will not draw on usage at all.
dax says stop shouting at the people who set the limits
dax pushed back on the reaction rather than the change. He says people do not understand how hard the OpenAI and Anthropic teams work to ship subscription products that at best break even while consuming limited compute, that nobody wants to cut limits, and that screaming at them does not help.
His second post is the more interesting one. If you grant the premise that burning a lot of tokens really is more productive, then the cost structure sorts companies by balance sheet, and early stage startups are the ones who cannot pay to play.
For working engineers
The useful takeaway is that the sticker price on one of these plans now tells you almost nothing. What matters is how the usage meter is computed, and that is the thing that just changed while the $200 stayed the same. Tibo's own framing points at where this ends, prices on the API dropping far enough that most people buy usage as needed and the gap between subscription value and API value stops mattering. Until then, the answer to "what does $200 get me" is a moving number that neither lab is obliged to hold still.
it's reverting back to the days when you had to raise money to buy servers to even launch