this post was submitted on 04 Jun 2026
532 points (99.1% liked)
Technology
85155 readers
4410 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- 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.
- Check for duplicates before posting, duplicates may be removed
- 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
view the rest of the comments
Yeah, what's the surprise here? Turns out it's expensive.
For the poor that will take the hit when the next bailout happens
The surprise is that they're not continuing to throw VC money at it to bolster user numbers and mislead investors.
Microsoft was deceptive here and never made it clear exactly what sort of deal you were getting with the flat rate. There was no indication of the actual magnitude.
Yeah especially because they now have some convoluted method involving different counts of processing, cache etc. But the developer has no easy way of seeing those statistics and thus has no feel for them. And developers already have little control over how much tokens a task takes. Which was fine with the flat rate, just use the service. But now that those things actually matter, the stats should be way easier to see?
So in typical Microsoft fashion not only did they raise prices they somehow made it even more shit. Like the AI already sucked, but does the service itself need to suck as well?
Not being able to control costs and very vague productivity improvement claims makes the ROI impossible to calculate. So even if the AI wasn't shit, it would still be hard to figure out if it's even helping at all.
Don't all the providers do this, though? Anthropic/Claude has different pricing based on if you're caching for five mins vs one hour (which are the only two options for cache TTL). https://platform.claude.com/docs/en/about-claude/pricing
Copilot was uniquely awful at this, because up to until literally days before the switch to usage based billing there was no way for people to track token usage, despite repeated calls from the community.
Microsoft only added a billing "projection" feature on the admin page that was meant to download a spreadsheet (which straight up didn't work for most people) less than a week before the new billing structure.
Ahh, that's not great. I've only ever used it for free since I get the $10/month plan for free as an open-source maintainer (https://docs.github.com/en/copilot/how-tos/copilot-on-github/set-up-copilot/enable-copilot/set-up-for-teachers-and-os-maintainers) so I've never had to deal with the billing side.
The amount of value they used to provide with the $10 plan felt too good to be true.
It was too good to be true. Microsoft had to implement additional 5 hour and week usage limits to cope with demand.
The problem was, for the entire history of copilot, there was literally no way to even check these usage limits. All usage counted towards these limits, even those with a 0x multiplier that didn't consume premium requests.
It was so bad that many people couldn't even use the premium requests they paid for the month. The only around it was to switch to auto model routing, which would tend to route to lower quality models.
Well, the point here is the deception. So, if you can find a similar link from the past from Microsoft…
You'd have to be truly ignorant to not know that it was going to happen (not saying that you are). It'll be interesting to see if there any legal recourse, but I'm guessing not.
All of these companies are hemmoraging money to train and provide LLMs. The only option is to charge more. This current increase, IMO, is an alpha test for future increases. Unless there is a major jump in the technology, moreso the logic to train LLMs more efficiently, the cost has to keep going up to stay solvent for most of these major players.
I’m not talking about ignorance here, just one of the facts about how Microsoft was being shitty.
Are you defending Microsoft’s choice to be shitty? Or just excited to share how you knew better?
Absolutely not defending Microsoft. I'm just saying that it was obvious even a couple of years ago that this current pricing model isn't feasible. It was always part of the plan to jack up prices, because that's how every subscription service operates. In the case of AI, the expenses are astronomical and the pricing has never been close to something resembling profit.
Microsoft announced these price increases back in April, this shouldn't be a surprise to anybody.
It was very much a red flag when they shut down signups.
It was easy to know how much the slop machine costs to run if you bothered to put even a tiny effort into finding out.