Gemini Advanced, the paid version of Google’s AI chatbot, costs $20 per month. That subscription fee gets you access to new and experimental AI models before they come to the standard Gemini experience, plus exclusive features like the ability to create Gems, custom chatbots that operate to your specifications.
This week, an APK teardown indicated that Gemini’s Deep Research function, currently exclusive to users with a Gemini Advanced subscription, will be coming to the free version in some capacity. Deep Research wouldn’t be the first paid Gemini feature to jump to the unpaid experience — other buzzy features like Gemini Live started paid before going free, too. It just doesn’t feel like there’s much reason to pay for Gemini these days.
Welcome to Compiler, your weekly digest of Google’s goings-on. I spend my days as Google Editor reading and writing about what Google’s up to across Android, Pixel, Gemini, and more, and talk about it all right here in this column. Here’s what’s been on my mind this week.
Deep Research, currently powered by the Gemini 1.5 Pro model and exclusive to the paid Gemini Advanced subscription, is a Gemini feature that can perform online research and assemble reports tailored to your specifications using what it finds. To be clear, Google hasn’t announced that Deep Research is going to be free. But Android Authority published an APK teardown earlier this week that suggests a “freemium” version of the functionality is coming to the base Gemini experience in the near future.
Android Authority puts “freemium” in quotes, which reads like the code reviewed by Android tinkerer AssembleDebug uses that word. If that’s the case, it does seem like the standard Gemini experience will be getting a version of Deep Research, rather than full access to the same feature that’s currently available with Advanced. It’s possible this free version of Deep Research could be limited to creating reports from a smaller number of sources, or that free users will only be able to use Deep Research to generate a certain number of reports in a given window of time.
Even so, yet another formerly paid feature coming to the free version of Google Gemini, even in a limited capacity, further erodes the value proposition of Gemini Advanced. Several big-ticket Advanced features have made their way to the free experience over the past year.
Gemini Advanced features tend to trickle down
Gemini Live, which featured prominently in Google’s overly sentimental Super Bowl ad, lets you have a back-and-forth verbal interaction with Gemini in the style of a conversation. The feature debuted in August, and I was intrigued enough to start a Gemini Advanced trial to check it out. Once the novelty of having a mock-conversation with my phone wore off, though, I found there wasn’t anything in Gemini Live worth continuing my subscription for.
Even if Live had really hooked me, I never would have had to pay to use it anyway: Gemini Live made its way to the free Gemini experience in September, just a month after it landed on the $20-a-month Gemini Advanced subscription.
Related
Google’s ‘heartfelt’ Super Bowl ad is as cynical as it is schmaltzy
It’s sweet, but what’s it saying?
Google’s made a number of formerly premium Gemini Advanced features free in just the past few weeks.
Last May, Gemini Advanced gained the ability to analyze uploaded files of various types to, as Google put it at the time, “Unlock deeper insights from your documents and streamline your workflows.” As of last month, it’s free. Now paid and standard users alike can ask Gemini questions about PDFs and Word documents.
Days after Google opened up document analysis to all Gemini users, it did the same for an even more specialized feature. Beginning in late February, the free version of Gemini now includes Code Assist functionality. Code Assist actually launched in 2023 under the name Duet AI for Developers, but was eventually rebranded under the Gemini umbrella as Code Assist. This new free version isn’t quite as versatile as what you’ll get with Advanced — there’s no integration with Google Cloud services on the free tier, for example. But you get up to 180,000 code completions per month without paying a dime, which Google Senior Director of Product Management Ryan J. Salva says is enough for even the majority of “today’s most dedicated professional developers.”
Later that same week, Gemini’s Saved Info feature, which lets you give your account’s instance of Gemini information that it’ll remember and apply in future interactions, also came to the free tier. It had debuted on the paid Advanced subscription a few months earlier, in November.
Even setting Deep Research aside, this feels like a pattern: Google debuts a flashy new feature on the paid Gemini Advanced subscription, and then, after a period of weeks to months, offers the feature (or at least a version of it) to everyone, for free. That hasn’t been the case with every Advanced feature, of course, and Advanced will always offer early access to newer Gemini models — like the recently released, experimental Gemini 2.0 Flash Thinking model.
For individuals not paying for Advanced, this is a non-issue. If you’re a casual Gemini user, the version of the chatbot you interact with is regularly picking up new features, and you probably don’t see much reason to pony up cash each month. On the flip side, it seems like a lot of Advanced subscribers are getting a raw deal; timed exclusivity of increasingly niche features only goes so far. Unless you’re a hardcore AI booster building custom chatbots and pushing context window limits, the biggest benefit you’ll get paying for Gemini Advanced is probably the terabyte of Drive storage Google’s $20/month plan includes.
Does Google need us to spend money on Gemini Advanced?
The money Google’s bringing in from a cadre of passionate users paying $20 a month to access cutting-edge features could only be a drop in the bucket relative to what AI services like Gemini cost to operate. Google’s got a lot of irons in the fire and its financials are kind of inscrutable to someone like me, but AI competitors like OpenAI and Anthropic, which offer comparable products to Gemini in ChatGPT and Claude, are currently operating at a loss. According to documents reviewed by the New York Times in September, OpenAI was braced to lose about $5 billion in 2024. Anthropic’s 2024 wasn’t much better, though that company is offering sunny projections for 2027 and beyond.
Despite all evidence pointing to AI currently being unprofitable as an industry, corporate investment in AI is through the roof. Google parent Alphabet has disclosed plans to spend $75 billion on capital expenditures this year. Much of that spend will be dedicated to building out AI infrastructure.
The main advantage to Google in charging $20 per month for Advanced — an amount high enough to chase off casual users, but still attainable to many enthusiasts — may be in containing new, resource-intensive AI features to a small subset of the general Gemini user pool as Google gathers usage data and builds out AI capacity. The real money for AI providers lies in providing service to large organizations (including, in OpenAI’s case, the federal government) and responding to API calls from third-party software that connects to AI (think Siri’s ChatGPT integration).
For regular users, though, there’s not a lot of advantage to paying for Gemini right now. There are edge cases — small business owners who could stand to benefit from a specific feature that’s currently exclusive to Advanced, for example. But if you’re a casual observer, eyeing some Advanced feature or other and thinking it might be cool to play around with, there’s a good chance you’ll get the opportunity to do so on the free tier in the near future.
