Google’s publicly-available AI offering has come a long way from the days of Bard. While the Gemini Flash model, available for free at gemini.google.com, is quite capable, it often falls short of the capabilities of the most advanced models (GPT-4o or the Claude 3.5-Sonnet) from Google’s competitors.

Google’s answer to the competition is their top-tier model, Gemini Pro, which powers their Gemini “Advanced” experience. Access to Gemini Advanced, however, requires a Google One subscription , which costs $19.99/month in US or ₹1950/mo in India. But if you are not interested in other benefits of the Google One subscription, or only have occasional use of the most powerful model, there is a free and easy way: Google AI Studio

You may not have heard of Google AI Studio because it is primarily targeted towards developers, and is therefore not advertised as part of the regular product marketing. Yet, it has an easy-to-use interface, making it well-suited even for non-developers.

Have a look at the screenshot below for a peek at the experience:

As you can see, it has a pretty standard chat interface, plus a lot of controls on the right. You can play with these variables (Temperature, Top P, etc.) to fintune the behaviour of the model though the default settings work well for most cases.

Accessing Gemini Pro via AI Studio comes with the following benefits:

  • Multimodal reasoning and analysis over multiple images, PDFs, videos, audio files etc. (I haven’t personally tested the limits extensively, but didn’t encounter any issues with 10 PDFs in a single chat)
  • 2 million token long context window, equivalent to ~1000 pages of text, ~10 hours of audio, ~1 hours of video, ~30K lines of code!
  • Google Search-backed grounding of responses
  • No/pretty generous free limits. Google only mentions that AI Studio usage is free of charge, and doesn’t call out the limits specifically. If we go by Gemini API pricing , they allow 50 free requests per day, maximum of 2 requests per minute, via the API. These limits should be enough for most use cases. In my usage over the last few weeks, I have only occassionally run into rate limits.

Note that being able to use Gemini Pro via AI Studio is NOT a hack or workaround that Google may shut down any time soon. Providing free (controlled) access to the suite of Gemini models is part of Google’s deliberate strategy to make development of Gemini-backed AI apps easier. That said, Google will not talk about this in their general-public facing marketing material to avoid cannibalising sales of the Google One subscription. Also, the AI studio experience offers too many advanced controls than the average user needs, and risks coming across as overwhelming.

But why use Gemini Pro when I have ChatGPT?

I have a ChatGPT Plus subscription. Several times I run into issues extracting data from PDFs, or analysing images with GPT 4-o. Gemini Pro, however, has consistently performed well.

Also, when you are doing something creative or sightly complex stuff, it is better to have a multitude of the most powerful AI models handy—each of them have their own strengths.

You can try Google AI Studio here.