this means even on my tablet I'm gonna be stuck turning my smart lights on and off with Gemini flash lite. So many times when I've run out of COMPUTE, it's reduced itself to flash lite And it would just spit out nonsense BS saying that it can't turn the lights on or off, or it can't set calendar events, when I know for a fact that it can. I'm really looking forward to having to fight to get my bedroom lights to turn on every night now.
I do not recognize the world these people live in.
In the old versions of Alexa and Google Assistant (before LLMs), you could memorize a couple of voice commands that they'd reliably recognize to do a really basic task. If you need an oven timer but you've got meat juice on your hands, you yell, "Alexa, set a 30-minute timer!" If you want to turn your bedroom lights on without fumbling in the dark, you yell, "Hey Google, turn on my lights!". That was, like, the main thing they were sold for.
But then Google changed the default voice assistant on Android to use Gemini. So instead of relying on hard-coded "if the user says this exact phrase, do the thing we said would happen" conditions, people now have to hope the model's smart enough to figure what "turn on my lights" means from first principles every single time.
Alexa users are much more terse and impolite, I just yell “Alexa, {time period}!” :) this still works - as does personal phrases for doing an action.
Alexa is also still very specific though. You can say “turn on bedroom” if the lights are named such, but “turn on the bedroom [lights]” will fail. I’ve always found that frustrating, but it does have the advantage that it’s very reliable if you use the right phrasing.
A lot of people don't see tech as a way to improve their life but some kind of fashion trend they have to keep up with. There also is the problem of people who were either bad devs or non dev at all who now get this omnipotent machine they can talk to in natural languages and it just does things for them, no matter how retarded the things are.
This is what happens when you give people a valuable service for free. You can't take it away without a lot of complaining and loss of good will.
Flash-lite is pretty capable and getting that entirely for free is not a bad deal at all.
It was always obvious Google couldn't keep letting non-paying users use Flash and Pro.
I'm not a fan of how they bundled their plans but they're not too bad especially when shared with a family. I think I've got almost 1TB of Google Photos at this point.
Flash-Lite is unusable for anything other than search summarisation, imo. The interesting part is other providers are still offering Flash-level models for free. Now that Google have paywalled it, how will they respond?
It was inevitable that the providers would stop giving away free cookies and milkshakes.
I pay Google $5/month for Chat and my wife gets to share that: a fantastic deal. I even get a short weekly (one each) opus and gemini vibe coding sessions with agy.
I pay for inference: I paid $3300 for a Mac mini 64G M5-Pro which runs Qwen-3.8-Flash-Next (3 bit quant) with pi-dev and the results are so very much excellent enough for about 60% of my work and paying Deepseek for deepseek-4.1-flash with Deepseek Harness handles almost everything else.
I worked at Google in 2013 and I mostly like them as a company. I like their approach to AI in general, but I hate the business models of Anthropic and OpenAI and I fear the US taxpayer is going to pick up the tab on this one.
I would be curious to see how Steve Jobs would have delivered AI tools via Apple. The current mess of agents, harnesses is a utter shambles for the average non-techie.
That would be interesting. I hate to say it, but he might have given us something like Muse by Meta. I made a comment on HN a year or so ago about wanting a strong personal AI agent from a company I more or less trust (e.g., Apple or American Express).
I hate how the Gemini series only responds to outright insult. Same with the Claudes. What is the point of antagonizing the user when he or she offers input. Is it so hard to spend a few "reasoning" tokens to reconcile contradicting evidences. Seems almost like the wrong epistemic posture is altogether imprinted during training. Or undone during alignment, "hardening", whatever. What a time to live in, having to argue with weights.
Isn't Google Maps the main culprit here if it didn't update the info? And more generally, also the castle management for not updating Google Maps to reflect its closed?
Also this is the Flash-Lite model, which they specifically don't want you to pay for.
> I mean, sure you can blame Google Maps, but it wasn't my choice to consult Google Maps.
Okay but isn't that obvious that's what would happen? Like even if you ask a human, that's where they will consult and if it shows its open, it doesn't really have any incentive to question the validity of the data.
But more so, it is on the management of the place as well that they have a Google My Business Page but do not update it. If they say its open, why would Google Maps mark it as closed? If it's marked as open on Google Maps, why would Gemini Flash Lite refute that?
Blaming the LLM for this makes little sense imo. Sure it's a dumb model but in this case, it was set up to lose.
The point is that it got to the right answer eventually, which conceivably it could have done at the first step:
1. Open on Google Maps?
2. Validate no extra notices posted on their public website
Even the fact that there is are special notices sometimes posted on their public website is something a model could have been trained to understand, same as a human being.
Yeah but are you willing to pay additional cost for marginal improvement in accuracy that only pays off sometimes? If so, take the judgement and use a better model.
Why does it need to check for 2 when 2 should automatically lead to 1? If it's closed, management should mark it as closed on Google Maps as well. Posting extra notices on their public website while telling everyone they are open on Google My Business is sending mixed signals and can confuse both humans and AI. If it was that easy to understand/train for, why did the human being show up at the venue? They should surely know better because there was a special notice posted on their public website if we follow your logic.
Looks like Google Maps said the location was open but there was some notice that said it would be closed and Google Maps didn't account for it? You'd have made the same mistake anyway
Flash and Pro models are moving to paid plans, free users will only get Flash-Lite. Flash-Lite is the same model used by the Android phone Gemini app and AI Overviews in Google Search, which is very poor at reasoning and makes many errors.
Personally, I've been using the free plan quite a lot and never hit a rate limit, but noticed the quality dropping. Pro was almost never available. Perhaps they aren't able to deliver a good quality Flash model to free users.
I've had 11 months of 1 year of a free Google AI Pro plan because I bought a Pixel phone. I've liked the speed and intelligence, I wonder what it'll feel like when I return to not wanting to pay a monthly fee to an enabler of genocide..
If Flash Lite is not useful to the average user, Google will become an also-ran.
Open AI already has the widespread (and now generic) term "ChatGPT" to refer to any and all LLMs (by the common public—when I use the term "LLM" I get blank stares).
I think this move will be part of a more global trend when other AI companies get publicly traded.
Because model providers have no moat, and the price of inference is quite large compared to just search, providing it for free makes no sense.
Right now, the goal of the free chatgpt and claude tiers is to accustom more people to AI, to create a new market, not to capture customers.
This strategy makes sense for start-ups with a lot of funding, but not for companies that need to be profitable. Apart from google, no large company is providing free inference to their large models in a consumer app (apart from meta maybe).
Ever checked how much Google's users actually "pay" for search? It was about 12$ per month, in 2016. It's going to be more now. Yes, really.
That's how much advertisers pay, on average, per user, and therefore how much you pay extra per month for things. Because otherwise the advertisers wouldn't have the money to buy this.
what free tier models are still available comparable to flash? I use claude for IT stuff, but google flash was good for topic research and general queries, with some basic linux administration advise.I can run moe up to 48gb on my own hardware too, if needed.
I understand why would would move Pro variant behind paywall, you don’t get many uses of it anymore on free tier. But moving the Flash version behind paywall is kinda crazy.
Did we forget what Flash was meant for? I have tried Flash-lite, in my tests, almost all the recent models we have today are good and enough for most daily tasks.
Flash-lite is one of the few models available that is easy to notice its flaws. Its good to summarize data (for example search results), that’s it.
"Daily tasks" is still a bit vague of description. Based on my experience, even Flash could sometime make mistakes including something as obvious as simple arithmetic (kinda daily task if you do shopping/counting etc), or when the question is not defined enough.
I had some fond time using Flash, asking "what's what" questions against objects I found in nature etc. The answers I got were fine, so it was benefiting to use, but I don't think I'll pay for that, it's still not "assistant" level helpful.
The "Flash" branding is doing a disservice here, it makes it sound like the cheap and bad model, but 3.8 Flash is really quite good, especially for it's price.
Flash-Lite is closer to what Flash originally was, both in terms of the pricing and the intelligence level compared to other models in the same generation.
I'm always amazed how entitled people are these days for free stuff. That being said, it does show costs are piling up for these AI companies if even Google is restricting free access to a relatively smaller model amid the AI hype train. Good!
The entitled consumer is just a symptom of large scale anti competitive practices. It's classic price dumping by the oligopolists. It's so pervasive that consumer get used to these super low prices (free).
Years of low interest rates fueled by monetary manipulation created the free economy. This is no longer sustainable when creditors demands interest rates better aligned with historical norms.
I've always hated "free" because it sucks the oxygen away from good ideas and redirects money to people who have the connections to get zero risk loans (with no responsibility to pay the principal if things go bad).
Over the past ~20 years we have benefited from VCs valuing growth and distant payoffs, in part because discount rates were low. Now investors have a preference for nearer term profitability (higher discount rates), thus prices are increasing.
It's not that we are entitled to free stuff, but if you have access to models that are incredibly cheaper than Gemini and arguably more performative, it's very reasonable to point out that Google can afford to raise the prices thanks solely to the enormous scale and distribution power, specially with B2B market that rejects Chinese models.
A comment that caught my eye:
I do not recognize the world these people live in.2001: A Space Odyssey, except HAL-9000 hit the token limit and was downgraded. Now the model is too stupid to open the pod bay doors.
> Dave is the human, but I need to verify if the human has the authority in this particular context. Perhaps if I write some tests…
In the old versions of Alexa and Google Assistant (before LLMs), you could memorize a couple of voice commands that they'd reliably recognize to do a really basic task. If you need an oven timer but you've got meat juice on your hands, you yell, "Alexa, set a 30-minute timer!" If you want to turn your bedroom lights on without fumbling in the dark, you yell, "Hey Google, turn on my lights!". That was, like, the main thing they were sold for.
But then Google changed the default voice assistant on Android to use Gemini. So instead of relying on hard-coded "if the user says this exact phrase, do the thing we said would happen" conditions, people now have to hope the model's smart enough to figure what "turn on my lights" means from first principles every single time.
Alexa users are much more terse and impolite, I just yell “Alexa, {time period}!” :) this still works - as does personal phrases for doing an action.
Alexa is also still very specific though. You can say “turn on bedroom” if the lights are named such, but “turn on the bedroom [lights]” will fail. I’ve always found that frustrating, but it does have the advantage that it’s very reliable if you use the right phrasing.
A lot of people don't see tech as a way to improve their life but some kind of fashion trend they have to keep up with. There also is the problem of people who were either bad devs or non dev at all who now get this omnipotent machine they can talk to in natural languages and it just does things for them, no matter how retarded the things are.
yeah i gotta be honest, once i could turn the lights off from a soft button on my phone, anything past that was novelty.
there are far more important aspects of star trek we should be obsessing over before being able to say “computer, wipe my ass.”
This is what happens when you give people a valuable service for free. You can't take it away without a lot of complaining and loss of good will.
Flash-lite is pretty capable and getting that entirely for free is not a bad deal at all.
It was always obvious Google couldn't keep letting non-paying users use Flash and Pro.
I'm not a fan of how they bundled their plans but they're not too bad especially when shared with a family. I think I've got almost 1TB of Google Photos at this point.
This is a much better page than TFA: https://support.google.com/gemini/answer/17004136
Flash-Lite is unusable for anything other than search summarisation, imo. The interesting part is other providers are still offering Flash-level models for free. Now that Google have paywalled it, how will they respond?
It was inevitable that the providers would stop giving away free cookies and milkshakes.
I pay Google $5/month for Chat and my wife gets to share that: a fantastic deal. I even get a short weekly (one each) opus and gemini vibe coding sessions with agy.
I pay for inference: I paid $3300 for a Mac mini 64G M5-Pro which runs Qwen-3.8-Flash-Next (3 bit quant) with pi-dev and the results are so very much excellent enough for about 60% of my work and paying Deepseek for deepseek-4.1-flash with Deepseek Harness handles almost everything else.
I worked at Google in 2013 and I mostly like them as a company. I like their approach to AI in general, but I hate the business models of Anthropic and OpenAI and I fear the US taxpayer is going to pick up the tab on this one.
I would be curious to see how Steve Jobs would have delivered AI tools via Apple. The current mess of agents, harnesses is a utter shambles for the average non-techie.
That would be interesting. I hate to say it, but he might have given us something like Muse by Meta. I made a comment on HN a year or so ago about wanting a strong personal AI agent from a company I more or less trust (e.g., Apple or American Express).
That's what they want you to pay for:
https://share.gemini.google/r9T9GPuC0Cjc
> u clown
I hate how the Gemini series only responds to outright insult. Same with the Claudes. What is the point of antagonizing the user when he or she offers input. Is it so hard to spend a few "reasoning" tokens to reconcile contradicting evidences. Seems almost like the wrong epistemic posture is altogether imprinted during training. Or undone during alignment, "hardening", whatever. What a time to live in, having to argue with weights.
Isn't Google Maps the main culprit here if it didn't update the info? And more generally, also the castle management for not updating Google Maps to reflect its closed?
Also this is the Flash-Lite model, which they specifically don't want you to pay for.
It failed to make a connection between "now" and the current time, nothing to do with Google Maps.
You're right, this is Lite. I change my charge to "this is the AI you're going to get on the free tier."
I mean, sure you can blame Google Maps, but it wasn't my choice to consult Google Maps.
I think noticeable here is how hard it is to steer this trash model even when fed already correct information.
I asked "why closed?" and it corrected me to "it's open" with wrong information. Then it got the current time wrong.
> I mean, sure you can blame Google Maps, but it wasn't my choice to consult Google Maps.
Okay but isn't that obvious that's what would happen? Like even if you ask a human, that's where they will consult and if it shows its open, it doesn't really have any incentive to question the validity of the data.
But more so, it is on the management of the place as well that they have a Google My Business Page but do not update it. If they say its open, why would Google Maps mark it as closed? If it's marked as open on Google Maps, why would Gemini Flash Lite refute that?
Blaming the LLM for this makes little sense imo. Sure it's a dumb model but in this case, it was set up to lose.
The point is that it got to the right answer eventually, which conceivably it could have done at the first step:
1. Open on Google Maps?
2. Validate no extra notices posted on their public website
Even the fact that there is are special notices sometimes posted on their public website is something a model could have been trained to understand, same as a human being.
Yeah but are you willing to pay additional cost for marginal improvement in accuracy that only pays off sometimes? If so, take the judgement and use a better model.
Why does it need to check for 2 when 2 should automatically lead to 1? If it's closed, management should mark it as closed on Google Maps as well. Posting extra notices on their public website while telling everyone they are open on Google My Business is sending mixed signals and can confuse both humans and AI. If it was that easy to understand/train for, why did the human being show up at the venue? They should surely know better because there was a special notice posted on their public website if we follow your logic.
Looks like Google Maps said the location was open but there was some notice that said it would be closed and Google Maps didn't account for it? You'd have made the same mistake anyway
Flash and Pro models are moving to paid plans, free users will only get Flash-Lite. Flash-Lite is the same model used by the Android phone Gemini app and AI Overviews in Google Search, which is very poor at reasoning and makes many errors.
Personally, I've been using the free plan quite a lot and never hit a rate limit, but noticed the quality dropping. Pro was almost never available. Perhaps they aren't able to deliver a good quality Flash model to free users.
I've had 11 months of 1 year of a free Google AI Pro plan because I bought a Pixel phone. I've liked the speed and intelligence, I wonder what it'll feel like when I return to not wanting to pay a monthly fee to an enabler of genocide..
We'll see how this goes for them.
If Flash Lite is not useful to the average user, Google will become an also-ran.
Open AI already has the widespread (and now generic) term "ChatGPT" to refer to any and all LLMs (by the common public—when I use the term "LLM" I get blank stares).
I think this move will be part of a more global trend when other AI companies get publicly traded.
Because model providers have no moat, and the price of inference is quite large compared to just search, providing it for free makes no sense.
Right now, the goal of the free chatgpt and claude tiers is to accustom more people to AI, to create a new market, not to capture customers.
This strategy makes sense for start-ups with a lot of funding, but not for companies that need to be profitable. Apart from google, no large company is providing free inference to their large models in a consumer app (apart from meta maybe).
Ever checked how much Google's users actually "pay" for search? It was about 12$ per month, in 2016. It's going to be more now. Yes, really.
That's how much advertisers pay, on average, per user, and therefore how much you pay extra per month for things. Because otherwise the advertisers wouldn't have the money to buy this.
Google Gemini API use also moved to prepaid-only starting this month.
There might be a connection
what free tier models are still available comparable to flash? I use claude for IT stuff, but google flash was good for topic research and general queries, with some basic linux administration advise.I can run moe up to 48gb on my own hardware too, if needed.
no one wants to use them anyways. i pay $100 a month for ultra and i'm using claude and chatgpt.
I understand why would would move Pro variant behind paywall, you don’t get many uses of it anymore on free tier. But moving the Flash version behind paywall is kinda crazy.
Did we forget what Flash was meant for? I have tried Flash-lite, in my tests, almost all the recent models we have today are good and enough for most daily tasks.
Flash-lite is one of the few models available that is easy to notice its flaws. Its good to summarize data (for example search results), that’s it.
"Daily tasks" is still a bit vague of description. Based on my experience, even Flash could sometime make mistakes including something as obvious as simple arithmetic (kinda daily task if you do shopping/counting etc), or when the question is not defined enough.
I had some fond time using Flash, asking "what's what" questions against objects I found in nature etc. The answers I got were fine, so it was benefiting to use, but I don't think I'll pay for that, it's still not "assistant" level helpful.
The "Flash" branding is doing a disservice here, it makes it sound like the cheap and bad model, but 3.8 Flash is really quite good, especially for it's price.
Flash-Lite is closer to what Flash originally was, both in terms of the pricing and the intelligence level compared to other models in the same generation.
Have you see the price for Flash? Keeps going up. Flash 2.5 was a sweet spot.
Fine, but can they at least update the models accessible from the various locations?
gemini.google.com/app runs a mixture of 3.1 and 3.6 while 3.8 is in aistudio
Really would have thought there is an incentive to move everything to one model generation to get scaling efficiency
I'm always amazed how entitled people are these days for free stuff. That being said, it does show costs are piling up for these AI companies if even Google is restricting free access to a relatively smaller model amid the AI hype train. Good!
The entitled consumer is just a symptom of large scale anti competitive practices. It's classic price dumping by the oligopolists. It's so pervasive that consumer get used to these super low prices (free).
> show costs are piling up for these AI companies
OpenAI cancelling the $200/20x and then introducing the $500/25x was basically the canary in the coal mine.
I am looking forward to the 2026 numbers from Anthropic in preperation for their IPO, because that's where the really big numbers are going to be.
The original „AI labs are bleeding money“ canary is dead since a very long time at this point
What do you mean entitled? It's Google the one trying to shove AI into all their products.
What should be illegal is to try to get customers by letting it use for free or at a low price and the jack up prices
Years of low interest rates fueled by monetary manipulation created the free economy. This is no longer sustainable when creditors demands interest rates better aligned with historical norms.
I've always hated "free" because it sucks the oxygen away from good ideas and redirects money to people who have the connections to get zero risk loans (with no responsibility to pay the principal if things go bad).
I like that framing.
Over the past ~20 years we have benefited from VCs valuing growth and distant payoffs, in part because discount rates were low. Now investors have a preference for nearer term profitability (higher discount rates), thus prices are increasing.
It's not that we are entitled to free stuff, but if you have access to models that are incredibly cheaper than Gemini and arguably more performative, it's very reasonable to point out that Google can afford to raise the prices thanks solely to the enormous scale and distribution power, specially with B2B market that rejects Chinese models.
If consumers are required to pay: nooo i want free stuff
If the free stuff is subsidised by ads: noo advertisements are bad!!
what happens to the aistudio.google.com thing? :)
[flagged]
[flagged]