still showing $12.99/month and 200gb
Strange, maybe try clearing your browser cache?
I'm not signed up and there's a nice 3-month offer I'm seeing as well.
I also have the 3 month offer. I can't see whether you can cancel after 3 months or are you then signed up for a year?
@slipperypete: The yearly option is separate, says $7.99/month ongoing for me. If you want to keep it switching to yearly afterwards would be the best value.
@Hiphopopotamus: mine shows a different offer
Google AI Pro (5 TB)
A$32.99 but its $8.49/mo for 3 months
this would be great to try for 3 months but too much annually
@slipperypete: So I asked Gemini and it said
Yes, you can cancel at any time.
Google One subscriptions, including promotional offers like the AI Premium trials, are completely contract-free. If you sign up for the 3-month offer, you can cancel it whenever you like during that period.
I think I'm gonna give thos a go, it's $8.49 a month for 3 months then I will swap
Checked on my android phone in the Google One app it shows the new price.
Might consider once I really run out of the 200GB storage.
Good deal, but I am still on this free Gemini Pro 18 months plan, Went to India last Dec, got 1 month pre-paid plan from Jio (AU $5.2 or INR 349), and activated Gemini Pro plan.
Does this only work if you have a plan with Jio?
Yes, as its offered by Jio.
Have you got a link to how to sign up to this in Australia? Does it make sense?
You need to have a contact in India. They just need to enter your gmail address in JIO app, when activating this offer. That is it. Once the pro plan is activated, you can use it from anywhere in the world.
Also, no need to renew the plan. So basically $5.2 gives you 18 months of Gemini Pro access and the following benefits.
Gemini app: Get higher access to our most capable model Gemini 3, plus get Deep Research and unlock video generation with limited access to Veo 3 Fast
Image generation: higher limits on Nano Banana
Storage: 5 TB of total storage for Photos, Drive and Gmail
Flow: Higher access in our AI filmmaking tool to create cinematic scenes and stories, including limited access to Veo 3
Whisk: Higher access to image-to-video creation with Veo 3
Gemini Code Assist and Gemini CLI: Higher daily request limits in Gemini CLI and Gemini Code Assist IDE extensions
NotebookLM: Research and writing assistant with 5x more Audio Overviews, notebooks
Gemini in Gmail, Docs, Vids and more: Access Gemini directly in Google apps
do you need to maintain your plan, ie pay INR 349 every month or is it a one off?
See my comment above
Also, no need to renew the plan. So basically $5.2 gives you 18 months of Gemini Pro access and the following benefits.
How does this compare to chat gpt
Afaik, neither ChatGPT or Gemini have referred to themselves as MechaHitler, so they both beat Grok.
Comparison table according to ChatGPT, confirmed with Gemini, and true from my experience:
| ChatGPT | Gemini | Claude |
|---|---|---|
| Best overall balance of features | Best Google ecosystem integration | Best long-form writing and analysis |
| Excellent writing | Good writing | Excellent writing |
| Excellent coding | Very good coding | Excellent coding |
| Excellent reasoning | Very good reasoning | Excellent reasoning |
| Built-in image generation | Built-in image generation | More limited image generation |
| Advanced voice features | Advanced voice features | More limited voice features |
| Broad third-party ecosystem | Deep Gmail, Docs, Drive and Maps integration | Focused assistant experience |
| Great for general users, creators and developers | Great for Google Workspace users | Great for writers, researchers and analysts |
Honestly, it's a fast moving technology and they're constantly edging eachother out in different areas. I tend to use Claude professionally, Gemini for integration with my Google stuff, ChatGPT for personal research and suggestions when writing, and Copilot for quick questions when on my PC.
Username checks out
The fact they're a reddit user also checks out.
Big one for me was the fact that everyone in your Google household gets access to Gemini Plus
I switched over from ChatGTP Plus (early subscriber) to Claude Pro yesterday. Had Claude via work and found its research output and desktop features are far more reliable and polished. Even some tasks that require more than reading comprehension, like asked it to find nearby areas with open spaces, and Claude was able to do a visual search of maps (focused on my suburb) and find me large areas within developed lands that are currently vacant. Clause Code is at par (both pro and cons) against CGPT codex (although their Atlas browser is better UI than using Claude browser plugins) but Claude co-work is a game changer, for not just file tasks but also web tasks that compile things back to files. We are living in the future.
ChatGPT pales in comparison to Claude. Claude's output is far higher quality and its reasoning is miles ahead. I argue with it at least 80% less than what I did ChatGPT.
Having used Gemini a bit recently in a Google workspace context, the claim of 'deep integration' with Google products is laughable. The model is pretty good, but the integration has a long way to go
Be honest, did you use Claude to write this?
Be honest, did you use Claude to write this?
No. But, as per the post, I used ChatGPT to create the comparison table then ran it through Gemini. I then modified it slightly so that it better matched my opinion.
I primarily use Claude but I wouldn't blanket recommend it to everyone. It depends on your use case. If you're already heavily in the Google ecosystem and only want AI to supplement your existing tools (Calendar, Drive, Photos, etc) rather become a primary tool in itself, then Gemini is great. That's how I use it.
I also use Ollama but the people who would use Ollama don't need me to tell them about Ollama.
I am hearing that Claude co work is the most useful MS AI integration so far - embarrassing for Microsoft, but also probably relevant to this comparison I would have thought.
Overall, yes, Claude is the best… But what's best for you depends what you're using it for. Also depends on your computer. My laptop has an NPU that can run certain Copilot tasks locally/offline.
You can have a thinking machine that lies to you, or you can have MechaHitler's uncomfortable honesty. Choose one.
and what's the uncomfortable honesty?
@BestTechAdvisor: That the usual suspects don't complain that MechaHitler is wrong, merely heretical.
"ChatGPT is the best AI" - ChatGPT
ChatGPT is the most balanced overall and would probably suit the average user the best. Gemini is great for people who want to use AI to broadly organise their lives and for light productivity. If you consider NotebookLM as part of Gemini (which it is in this subscription) then it's the best for academics imo. Claude's definitely the best for professional power users, coders, etc. If you're a dev, then you start looking at things Ollama, or wrappers like OpenHuman or Hermes, etc, but most would still use Claude as the centrepiece or for complex reasoning.
Still, like I said, this all fast moving. This comment could be completely out of date in a matter of weeks.
I heard Claude hits the limit quite easily / quickly even on the paid subscription. Is that true?
So you pay them but you cant use them as much as you want and when you need it? how does that work?
Does it also work the same for ChatGPT paid plan (and other AI paid plans)?
Thinking to purchase an AI plan to explore the tech more but want to just get one plan.. sounds like it's either ChatGPT or Claude based on your comparison above.
I heard Claude hits the limit quite easily / quickly even on the paid subscription. Is that true?
It really depends on how heavy of a user you are. I'm a very heavy user and I do hit the limits quite often regardless of which AI platform I use.
So you pay them but you cant use them as much as you want and when you need it? how does that work?
It varies, but basically you pay a monthly subscription which includes a certain amount of usage, then you can pay for extra usage of you hit those limits.
Does it also work the same for ChatGPT paid plan (and other AI paid plans)?
Yes. It varies, but yes, all the major AI platforms have various limits, eg daily + monthly limits.
Thinking to purchase an AI plan to explore the tech more but want to just get one plan.. sounds like it's either ChatGPT or Claude based on your comparison above.
What are you hoping to use it for?
@BestTechAdvisor: do they tell you your daily and monthly limits? or it just depends on their capacity at the time, i.e. variable/dynamic rather than fixed limits?
what do you use AI paid plans for if I may know? personal or work stuff?
tbh, at this point I just want to explore the tech further. May do some vibe coding and most likely usage will be research-related stuff. I am also thinking of playing around with agentic AI via Hermes or OpenClaw, combining with local models like Qwen 3.6
do they tell you your daily and monthly limits? or it just depends on their capacity at the time, i.e. variable/dynamic rather than fixed limits?
On the paid plans, most will have fixed limits. When there is a surge of usage, paid subscriptions will be prioritised over free users.
what do you use AI paid plans for if I may know? personal or work stuff?
I use paid plans for both personal and work stuff. I'm a very heavy user and I'm not using AI for just about everything, e.g. planning travel, workout tracking, relationship manager, researching for articles, troubleshooting customer issues, etc, etc.
tbh, at this point I just want to explore the tech further. May do some vibe coding and most likely usage will be research-related stuff. I am also thinking of playing around with agentic AI via Hermes or OpenClaw, combining with local models like Qwen 3.6
If you're going to check-out Hermes or OpenClaw then Claude is the paid service I'd recommend. I'm playing with OpenHuman with Ollama at the moment.
they both beat Grok
Uninformed bias, effectively disinformation. Unsurprising from the self-declared "best tech advisor"!
Grok is a powerful LLM in all tasks. General chats including spoken interaction, research, image gen, and coding. It's childish to dismiss it via low-quality mud-slinging.
I'm a developer and I've used them all. I'm also a casual AI user for researching things in moderate depth (pages of deep diving a topic), creative tasks, and general knowledge. Grok does it all, and does it well.
I showed my elderly relative the Grok app, when the voice is in "Gork" mode, and they loved it. The Gork voice character is a nice balance of friendly, informative, and isn't afraid of playing a little sarcasm and backchat without being offensive. The balance is actually very well done. This is SuperGrok I'm talking about.
@BestTechAdvisor: Not only won't I take your word for it, I won't take your links for it either. I also don't trust "trust pilot". What? But they have "trust" in their name? LOL.
I'm reminded of the cheap corner store in middle of nowhere proudly declaring "world's best coffee". A guaranteed signal the coffee will be bad.
@cerealJay: Ok. You don't have to trust Antutu, 3DMark, etc either, but it'll be hard arguing against industry standard benchmarks.
industry standard benchmarks
For an industry in infancy, your attempt to align with 30 year old "3DMark" is laughable.
The point is, your advice is poor if the single criticism you have for a major player like Grok, is a dopey spat about "mecha-hitler" which sounds like it came from the arse of reddit.
In Cursor, I often leave the model choice on Grok because it does a good job. I wouldn't do that for any reason other than it's a good choice that delivers good results. Meanwhile, your advice is "something something mecha-hitler". LOL.
For an industry in infancy, your attempt to align with 30 year old "3DMark" is laughable.
Arena is valued at over a $1 billion in an industry worth hundreds of billions, but you don't think they can put an accurate benchmark together?
What's your evidence that Grok outperforms Gemini and ChatGPT?
your advice is poor if the single criticism you have for a major player like Grok, is a dopey spat about "mecha-hitler"
It's not my only criticism - I just gave you the industry benchmark that breaks down performance in various areas - but it's a shining example of Grok's well publicised deficiencies.
Any authority on AI will agree with me: xAI is lagging DeepMind, OpenAI, and Anthropic. If you want a concession, xAI is beating Microsoft and Meta, but that's not really saying much.
@BestTechAdvisor: You're fixated on "benchmarks" that don't apply like they do for 3D graphics. We all reach the limits of 3D graphics by simply playing the same games. That's why 3D benchmarks apply equally to everyone who play the same games.
Not so with AI use. We are not all "playing the same game." The vast majority of us are not shredding their AI prompts with colossal repos or task lists. The quality of prompts would vary a lot. For example "make a cinematic masterpiece from this doodle" is the kind of dumb beginner prompt that nobody wants a benchmark for.
You linked to "Agent Arena", a service launched in June 2026 from a commercial player, then try to argue "you can't argue with benchmarks"!
Grok's well publicised deficiencies.
That you conveniently didn't re-publish here, other than something you heard one time.
Cursor, the hugely popular coding platform, recently described Grok 4.5 as "their most intelligent model". They go on to say "Grok 4.5 can handle difficult, long-running tasks that require creatively using tools to solve problems, whether in software engineering, data science, finance, legal work, or anything else you do on a computer."
But nevermind direct quotes from major coding platforms, if you prefer to take comfort in your sketchy Leaderboard or the vaporous "well publicised deficiencies" then good luck to you.
@cerealJay: You waited weeks before responding the first time and now you've waited another 4 days before responding.
Even with all that time you still haven't come up with any sort of benchmark.
Can I expect timely responses with some tangible evidence going forward or should I let you sporadically yell at clouds?
waited another 4 days
LOL what? Casually changing topics to "comment response time"! Is that your cunning plan? LOL you got me.
sporadically yell at clouds
Sounds like something AI would spit out. Was your prompt "getting owned in comments, I need a comeback that throws mud and asserts my tech authority"?
As I said, your linked benchmark service was launched last month, tied to a commercial product. Hardly an "industry reference". I just finished telling you what the Cursor team said the other day about Grok 4.5, so the ball's still in your court. Not my fault if you're fumbling about like under 13 soccer.
LOL what? Casually changing topics to "comment response time"! Is that your cunning plan? LOL you got me.
1) You converse like we're on Xbox Live.
2) I'm not casually changing topics, I'm pointing out that you take an extremely long and varying time to respond and the quality of your responses are poor. You don't provide evidence and your claims are false.
Sounds like something AI would spit out. Was your prompt "getting owned in comments, I need a comeback that throws mud and asserts my tech authority"?
You're not familiar with the popular meme about yelling at clouds?
As I said, your linked benchmark service was launched last month
No, the benchmark launched in 2023 and is the most respected benchmark in the industry. Yet more evidence of the Dunning Kruger Effect…
Can I see your benchmarking now? I keep asking and you don't provide anything.
I just finished telling you what the Cursor team said
Cursor is owned by xAI, of course they're going to claim that xAI is the best AI.
@BestTechAdvisor: You lost the plot from the start by answering a question nobody asked.
This is a bargain thread about Google AI Plus. Someone asked how it compares to chatGPT, and you answered with "mecha-hitler Grok". The burden is yours to provide evidence of your rubbish claims about Cursor's most intelligent model. All you've done is provide a link to a leaderboard who literally added Grok 4.5 to its table yesterday, July 13 2026.
No, the benchmark launched in 2023
Their own changelog says:
June 4, 2026
The Agent Arena leaderboard is now live. It ranks models on real-world agentic evals at scale, measuring behavioral signals like file downloads, disapproval events, retries, and steerability rather than static benchmark scores or preference votes alone. Check out the Agent Arena blog post to learn more about the methodology.
The link you posted to the leaderboard has no archive in Wayback prior to March 2026. At this point, you need to stop dribbling about your leaderboard, it's irrelevant.
People here want to know how these AI services compare for a range of tasks. Have you noticed I'm not trashing the other models? They all have uses and strengths. Your "best tech advice" was on shaky ground, now in pieces on the floor.
of course they're going to claim that xAI is the best AI.
Oh sure, because they just lie to all their users!
OR… it's true: Grok 4.5 is their strongest model to date because it was trained on a broader data mix: trillions of tokens from Cursor's real developer usage + high-quality STEM tasks, research papers and general knowledge. This makes it more capable beyond pure coding.
Are you done? Hopefully you're done.
@cerealJay: For anyone else who may be reading, this will be my last reply to this thread (and to @cerealJay in general). In case anyone stumbles across this conversation, I thought it was worthwhile making it clear that this person shouldn't be listened to.
I've shown this before, but cerealJay keeps kicking the can down the road, so I'll show it again, decisively.
I personally think cerealJay is being deceitful on purpose. If cerealJay is simply being ignorant, this is because they suffer from the Dunning-Kruger effect, whereby they know just enough about a topic to think they know a lot, but not enough about the topic to realise how little they actually know. Outsourcing your thinking to AI can worsen this effect.
Claim: "your linked benchmark service was launched last month"
This simply isn't true. The benchmarking service, with leaderboard, has existed for years. You can see an earlier iteration here, where the last update is in 2024.
cerealJay might be confusing the wider benchmarking service with the new iteration of the Agent Arena Leaderboard that uses new methodology, which is linked in the changelog that cerealJay later refers to. The reason I suspect cerealJay is being intentionally deceptive (and not just ignorant) is that cerealJay doesn't bother to link to the changelog, which clearly shows the leaderboard change history back to July 14, 2025, which states "We made improvements …", proving there is clearly history even before that.
At best, cerealJay is reading their own 'evidence' to ensure it's fit for use. At worst, cerealJay is being intentionally dishonest and is a bad actor.
Claim: Agent Arena is a "commercial player" [implied bias], Cursor is less biased and more trusted due to being a popular coding platform [implied neutrality].
Agent Arena is now a "commercial player", but it was originally launched by researchers from UC Berkeley and is Open Source. You can even fork it on GitHub. For this reason, it is difficult/impossible for Agent Arena to be meaningfully biased, and this is one reason they're the most highly trusted AI benchmarking platform.
Arena also have no obvious profit incentive to be biased. In fact, now that they are for-profit, their profit incentive is in accurately evaluating AI models.
Cursor is the opposite. Cursor is owned by xAI (Grok) and has a profit incentive and top-down pressure to be biased towards xAI. For some reason, cerealJay thinks it's ludicrous that companies will say that their product is the best, as if Elon Musk hasn't repeatedly claimed that "Tesla make the best cars".
Claim: Cursor [implied neutrality] says Grok 4.5 is their "most intelligent model"
This is another example that makes me think cerealJay is an intentionally dishonest liar and not just innocently ignorant. cerealJay has cherry-picked pieces of quotes from a public statement made by Cursor alongside xAI (their owner).
The full quote is "Today we are releasing Grok 4.5 together with SpaceXAI, our most intelligent model and the first we've built for more than software engineering."
That's right: cerealJay has edited the quote so it doesn't show that Cursor is releasing the AI model they've built with xAI. When Cursor says it's "our most intelligent model", they mean that Grok 4.5 is the most intelligent model that Cursor and xAI have released, NOT that it's the most intelligent model you can use via the Cursor platform.
cerealJay claims that the most-trusted, open-source benchmarking platform shouldn't be trusted as it was only launched last month (it was actually launched years ago) and is commercial (was originally not-for-profit, now profits from evaluating AI models).
cerealJay hasn't provided any evidence of their own; instead, cerealJay says we should trust xAI (owned by Elon Musk) and their subsidiary when they claim their AI model is the "most intelligent".
I've thoroughly debunked cerealJay. I've brought receipts. If cerealJay replies, you can't assume cerealJay is arguing in good faith. DYOR - research anything they (mis)quote. I've linked to a widely respected, open-source benchmark - that is the standard Grok fans must meet.
"…childish wall of text…"
Shifting your focus to "cerealJay" is kinda unhinged.
This started when you unhelpfully said ChatGPT and Gemini beat Grok because "mecha-hitler". People seeking tech advice aren't interested in your mud-slinging bias.
I provided balance by pointing out Grok is a proven highly capable model for a range of tasks. I did not trash-talk Gemini or ChatGPT. In professional coding tasks I've had better results with Grok. Others have different results. That's fine.
Notice how I'm not dribbling like an emotional child as I calmly talk about this topic?
The fact Grok 4.5 was only recently added to your leaderboard, obviously means it won't rank on the leaderboard. Other benchmarking has Grok 4.5 performing well in these early days. It does well in cost and speed.
The multiple leaderboards associated with your link is confusing and messy. People want to know if the AI can do good images, chats and research, good code etc and how it compares. You didn't meet that challenge when you trash-talked one of the best models because "Musk bad man". I've called you out, you throw tantrum. That happened.
… they mean that Grok 4.5 is the most intelligent model that Cursor and xAI have released, NOT that it's the most intelligent model you can use via the Cursor platform.
I never said otherwise. You make things up about what I said, then unleash name-calling. "Liar" etc, so childish.
Even though Cursor means the most intelligent model they have released, it means Grok 4.5 beats Composer and others. Grok 4.5 has trained on millions of developers using Cursor, which is a good qualifier for "most intelligent" as an absolute in that context. Grok 4.5 is here and performing at top tier level. Your trashy opinion of Grok can be safely disregarded as dopey bias.
I've thoroughly debunked cerealJay.
LOL, you win the least intelligent ranking on the most trusted industry leaderboard.
Gemini has been very helpful for me so far. I've won some cases with Gemini's help. I haven't won any claims with ChatGPT. Also, personally, Gemini has helped me fix some coding issues with Liquid files. ChatGPT either couldn't help or just gave me random code that didn't work.
Currently paying $70ish, will move this once the Turkish geopricing stops working.
Turkish pricing will likely change too
I was always a stinge, but allowing my entire family to automatically back up their photos is a god send. If times comes will pay the $$
Interesting. But if you suddenly die will your family ever be able to recover their precious memories.
@Snoovey: It's easy enough to leave people you trust with enough details to get in and download it all if needed.
@johndoh89: and how many people actually think of doing that. Plus it's hard to do even if you know you are going to die. And also who can you trust. Why neg such a common sense observation
OneDrive Turkey family plan for just storage, this is more for the AI benefits.
TRY23.99/mo (~$1/mo) for 6 accounts. Requires sign up via the OneDrive iOS app
Edit: You mention below you have it too :p
interesting it is still working for you. they had a crack down on a lot of us and we could not continue unless we had a Turkish payment method
There's still a fair few on reddit that have it. I even used a credit card and it worked. I had to use TurGame for my onedrive subscription. I am over the free limit so I cant receive emails if I leave microsoft 365. I am stuck paying…
Did it allow a card directly on Google or the app store? It's rediculously hard to fool Google for region hacks, but once you're signed up I haven't heard of any major issues. Have heard of the odd cancelled subscription renewals (with remaining time left) and conversion to local pricing upon renewal; standard practice for if you move countries. They force you to update your billing profile.
Same, think year 2tb was 2049 try or so.
This is using an old iPhone 11 connected 24/7 with Turkish vpn.
is that so they don't pick you up on a VPn. Maybe I should do that with the primary acc…
@R3XNebular: Yeah. Don't know if makes any difference as most account used here in Australia without VPN. But not gonna touch it. It works for me. Also same vpn for my Xbox sub and works fine.
Didn't know they had annual plan for 2TB, it's listed as TRY 204.99/mo on app store API.
For those who want ultra at all, the below countries bill in USD and it's US$199.99/mo / A$283.67
Antigua and Barbuda, Belize, Bermuda, British Virgin Islands, Cape Verde, Cayman Islands, Fiji, Mali, Niger, Palau, Papua New Guinea, Seychelles, Sierra Leone, Tajikistan, Turkmenistan
May be possible to use a normal card, if not USD iTunes cards are purchasable directly at cost on the Apple website. Various other AI apps have switched to USD pricing for countries with volatile currencies, Fiji is the go-to usually. The price is still reduced heavily vs the actual US pricing, just the currency is USD so they don't need to consistently update it.
Same here. Paid around 62 just last month. 1-year sub and now double the storage. So good so far!
https://i.imgur.com/81WHzgz.png
I'm seeing a 16% discount on the annual plan, so $79.99/yr for 400GB + AI Plus.
Tempting but I really don't use AI enough to justify going beyond the Basic plan that has no AI features.
These discounts are targeted, it's near random - I switch accounts and each one varies. There's also targeted sign up bonuses, up to half price for a year, a discount for 3 months, or a month free.
79.99 AUD Annual.
I was on a 2TB non-AI plan and today received an email from Google informing me that I’ve been upgraded to AI Plus while retaining my 2TB storage.
They added the AI stuff to the 2TB Google One plan a while back, now they're just changing the name.
I'm on the 2TB plan also (now with AI Plus) but, it doesn't show me an option to downgrade to AI Plus with the 400GB.
Edit: Went via another page and is now showing the option to downgrade to the cheaper plan.
And you want to? Was it supposed to able to? I don't use this stuff so am unaware how the membership works.
@Hiphopopotamus: Yeah, when I went above 250GB of storage (mostly Google Photos) the next tier was 2TB, at $149.99. This change adds an inbetween tier, and thus a saving.
@byrno: How do you downgrade?
When I clicked downgrade, it shows the subscription of the "downgraded plan" will start at the anniversary date later this year??
And it wants me to put down payment option again??
Confused
what was the link? I dont see an option for the 400GB plan. Im also on the 2TB
400 GB total storage
What is this? Storage for ants ?
he's absolutely right
One fire ant is 1gb of data and there's more than 400 to a colony so it can't be!
Claude suggests a fire ants DNA is only between 0.15 and 0.5GB. And assuming there will be little genetic difference between ants in a colony 400GB should do. Then you'll need a few data centres to evolve that DNA into a small model on someone's screen. We are no where near replicating a living colony yet. Not from DNA upwards
400GB is a fair bit, what are you storing JV??
Linux distros.
I store those on my free 125TB Office 365 developer storage.
Many distros, so many distros.
This seems like a strange strategy to drop prices. I use YT premium lite and ChatGPT and was debating whether to get Google Ai pro at $33/mth as GPT+YT premium lite is $40/mth and previously Ai plus + Yt premium would be ~$25 so an extra $8 for pro seemed worth it. Now I’m thinking maybe just get AI plus & YT premium lite for ~$18
It's very clever.
You drop prices, increase storage and introduce AI. More people will consider it and as a by-product they can scan all the files stored there and get better/more information about individuals.
I think it is more so to drop the price because they need more people to quickly sign up. They can then start pushing up the price yearly and people will have no option to leave because they have now used up so much storage.
That may be also the case as it may become difficult to migrate such a high volume of data, however being google its definitely for data mining. Same way they introduced AI to emails and summaries etc.
AI Plus is the new budget offering - it includes a lot less than pro.
No integration with other Google services (except Gmail), 2x instead of 4x usage, no access to pro models or deep research, etc. I don't see it as much of a benefit over a free plan - it's only double the usage and free plans can use Flash anyway. The Gmail integration is the main benefit - it's a good price for that alone. It can summarise emails, formalise, write for you based on a prompt, etc. Scans past emails and the email chain for context automatically (no data stored afaik). Will reference specific past emails when formalising (dates included) so you don't have to spend ages digging. Very helpful if you do a lot of formal emailing (ie I have an Administrative Reviews Tribunal (ART) matter, absolute lifesaver for this).
Was $12.99/Month and 200GB.