this post was submitted on 16 Jun 2026
180 points (99.5% liked)

Technology

85515 readers
4301 users here now

This is a most excellent place for technology news and articles.


Our Rules


  1. Follow the lemmy.world rules.
  2. Only tech related news or articles.
  3. Be excellent to each other!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
  9. Check for duplicates before posting, duplicates may be removed
  10. Accounts 7 days and younger will have their posts automatically removed.

Approved Bots


founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] Blue_Morpho@lemmy.world 21 points 1 day ago (3 children)

There was an article yesterday that if you sign up for the most expensive OpenAI plan and actually use it fully, OpenAI loses $14k a year on you.

This reminds me of the early Internet days. I ran an ISP. Like everyone else we offered "unlimited" service. But if everyone actually used unlimited we would have been bankrupted because it would tie up a phone line (channel of a pri) and modem on the portmaster. Regular users only used 1/10th of "unlimited".

[–] BlackLaZoR@lemmy.world 1 points 15 hours ago (1 children)

I think that number is massively inflated. $1k loss is much more reasonable value

[–] Blue_Morpho@lemmy.world 1 points 10 hours ago (1 children)

Where did you get $1k? Semianalysis's report is paywalled but seems thorough.

[–] BlackLaZoR@lemmy.world 1 points 10 hours ago* (last edited 10 hours ago) (1 children)

Rough estimation. Reasoning + answer is ~1k tokens. API price for Chat GPT 5.5 is $30 per 1mil of output tokens, $5 per mil, for new input tokens and $0.50 for cached tokens. They allow 3k reasoning prompts per week.

Quick math 12000prompts x1000tokens=12mil tokens per month which will be 12x30= $360 add all long context costs at and you can double or triple that cost.

If you consider non reasoning prompting cost can be scaled to infinity since as far I'm aware subscription offers unlimited access for that. But they have anti abuse clause in the agreement, so I doubt you'll get to anywhere close to $14k cost

[–] Blue_Morpho@lemmy.world 2 points 9 hours ago (1 children)

They allow 3k reasoning prompts per week.

It's their unlimited plan that loses them money. Where did you get 3,000/week? Their message cap is hidden and users complain they don't know how much of their "unlimited" is used up before being cut off.

[–] BlackLaZoR@lemmy.world 2 points 9 hours ago* (last edited 9 hours ago)

They forgot to update polish translated help site. What a clusterfuck.

https://help.openai.com/pl-pl/articles/11909943-gpt-55-in-chatgpt

Użycie Thinking

Jeśli korzystasz z planu Plus lub Business, możesz ręcznie wybrać GPT-5.5 Thinking w selektorze modeli z limitem użycia do 3000 wiadomości tygodniowo. Po osiągnięciu tygodniowego limitu zobaczysz wyskakujące powiadomienie, a GPT-5.5 Thinking nie będzie już można wybrać z menu. Jeśli korzystasz z planu Go, możesz wybrać Thinking z menu narzędzi, klikając ikonę + w polu wpisywania czatu. Użytkownicy Go mogą wysłać do 10 wiadomości co 5 godzin po włączeniu Thinking. Automatyczne przełączanie z GPT-5.5 Instant na GPT-5.5 Thinking nie wlicza się do tego tygodniowego limitu, a ChatGPT może nadal przełączyć się na GPT-5.5 Thinking nawet po jego osiągnięciu.

This paragraph doesn't exist in eng version

Right now it says Plus and Business plans are limited to 3000 messages per week for thinking, and that automatic switch to thinking doesn't count towards limit.

Thy updated yesterday apparently so who knows what was written a week ago... So I don't know, If I misread business for pro, or did they update policy again and again.

Edit: Correction eng exists but says

If you are on Plus or Business, you can manually select GPT-5.5 Thinking from the model picker. If you are on a Go plan, select + in the chat entry box, then select Thinking. Go users can send up to 10 messages every 5 hours after enabling Thinking. Automatic switching from GPT-5.5 Instant to GPT-5.5 Thinking does not count toward manual Thinking usage limits.

So they have some unspecified limit?

[–] GalacticRobot@lemmy.world 8 points 1 day ago

Isn't that basically every 'all you can eat' model though? 20% of people will fully max their usage, causing a loss, but the 80% of users will use far less than the purchase price, netting $$$ for the company. When that balance shifts, most places just put more limits to that 'all you can eat' or raise prices. Sure, I have unlimited internet, but it's difficult to symmetrically max out my 5gbit connection to get that full 'cost' each month.

As long as OpenAI has a health amount of money in the bank ($50 billion it seems), they have a long time to burn money, and then you just keep getting people to invest. Pretty sure that's been the model of every tech company for a long time. I think Uber took almost 15 years to be profitable, and had almost $50 billion invested. For a company that owns essentially zero cars, pretty wild.

[–] XLE@piefed.social 8 points 1 day ago* (last edited 1 day ago)

I take your ISP didn't have a free tier that also lost money, though, unlike these companies with apparently infinite money to give away