Google Bird is no more. Almost exactly a year after first introducing OpenAI's (hasty) effort to counter ChatGPT, the company has retired that name and rebranded Bard to Gemini, the name of its underlying model family. But more importantly, we're also releasing Gemini Ultra, our most capable large-scale language model to date.
However, Gemini Ultra is a paid experience. Google is making this feature available through its new $20 Google One tier (with a two-month free trial). This includes 2 TB of storage and the rest of the feature set of Google's One, including access to Gemini in Google Workspace apps such as Docs. , Slide, Sheet, Meet. Along with this, Google will also be discontinuing the Duet AI branding, which was mostly limited to Workspace's AI capabilities, and moving that to Gemini as well.
The company also released a new Gemini app for Android and plans to bring it to the Google app on iOS. On Android, you can now also replace Google Assistant with Gemini. For developers who want API access to the Ultra model, Google said it will have more to share in the coming weeks.
Gemini Advanced will be available in over 150 countries and territories, but will be available in English only for now. Next on the roadmap for additional languages are Japanese and Korean.
Image credits: Google
Gemini Ultra 1.0
Image credits: Google
When Google first announced Gemini, only the Gemini Pro model was widely available through Bart. At the time, Google said that Gemini Pro's performance was about on par with his GPT-3.5, but with GPT-4 now widely available, the announcement felt a little underwhelming. At the time, Google announced that its flagship Gemini Ultra would be available to consumers in early 2024 after a series of private tests. Although the company didn't explicitly say so, the understanding at the time was that the Ultra model would be part of a paid plan then called Bart Advanced, now renamed Gemini Advanced.
“Gemini Ultra 1.0 is a model that sets the state-of-the-art across a wide range of benchmarks spanning text, images, audio, and video,” Google's Sissy Haxiao said in a press conference ahead of today's announcement. “For Google, Gemini is more than just a model. It's about cutting-edge technology and building on top of it, from products that impact billions of users to the APIs and platforms that developers and enterprises use to innovate. It’s a shift in thinking about the whole ecosystem that we’re building.”
Image credits: Google
Haxiao also noted that Google decided to change its name to Gemini to reflect that the company's cutting-edge technology is at the heart of Bart (although it does have a large amount of branding expertise). If a savvy Google decided to launch a more capable model and wanted to call it Aries, instead of Gemini Ultra Super Pro Max 3.0, it would be called Bard). And while Google promotes Bard's scores to its users, many early users never returned after getting admittedly middling results, making the rebrand almost mandatory.
Bard had already switched to Gemini Pro, so there won't be any major changes for free users. However, if you choose to pay for Gemini Advanced, you will have access to the Gemini Ultra 1.0 model. You'll have to try it out for yourself to see how good Gemini Ultra 1.0 really is. Google itself was pretty vague about the feature at a press conference this week.
“With access to our Ultra 1.0 model, Gemini Advanced is capable of highly complex tasks using a variety of applications, including coding, logical reasoning, following subtle instructions, and creative collaboration. It has improved significantly,” Haxiao explained. “With Gemini Advanced, you can not only create longer prompts, but also better understand the context of previous prompts.” It's worth noting that Gemini Ultra 1.0 is also multi-model. This allows you to participate in conversations about images, for example.
gemini advanced
As for Gemini Advanced, currently the only way to access Ultra 1.0, users will have to sign up for the new $20 Google One AI Premium plan. This plan includes all the benefits of the existing Google One premium plan: 2 TB of storage, exclusive Google Photo editing features, premium video features in Google Meet, and enhanced event scheduling in Google Calendar. Additionally, you'll have access to more advanced models wherever Gemini is available. You'll soon be able to access Gemini from Google Workspace as well, presumably in a Microsoft Copilot-like experience, but Google didn't want to reveal any details about this just yet.
Image credits: Google
Currently, the current price for cutting-edge AI chat tools, including ChatGPT, is $20 per month. Google has the advantage here of being able to layer on these additional features without much additional cost (after all, the margins on these existing Google One storage plans are probably already pretty high) .
There is one nice perk. Even if you're already a Google One subscriber, you can also get a two-month free trial when you upgrade to the new AI Premium plan.