Google launches Gemini Ultra, its most powerful LLM yet

Google Bard is no more. Almost exactly a year after first introducing its (rushed) efforts to challenge OpenAI’s ChatGPT, the company is retiring the name and rebranding Bard as Gemini, the name of its family of foundation models. More importantly, though, it is also now launching Gemini Ultra, its most capable large language model yet.

Gemini Ultra will be a paid experience, though. Google is making it available through a new $20 Google One tier (with a two-month free trial) that also includes 2TB of storage and the rest of Google One’s feature set, as well as access to Gemini in Google Workspace apps like Docs, Slides, Sheets and Meet. With that, Google will also sunset the Duet AI brand, which was mostly restricted to its AI features in Workspace, and move that to Gemini as well.

The company is also launching a new Gemini app for Android and bringing it to the Google app on iOS — and on Android, you can now replace your Google Assistant with Gemini, too. As for developers who want API access to the Ultra model, Google said it’ll have more to share in the coming weeks.

Gemini Advanced will be available in more than 150 countries and territories, but only in English for the time being. Japanese and Korean are next on the roadmap for additional languages.

Gemini Ultra 1.0

When Google announced Gemini, it only made the Gemini Pro model widely available through Bard. Gemini Pro, Google said at the time, performed at roughly the level of GPT-3.5, but with GPT-4 widely available, that announcement felt a bit underwhelming. Back then, Google said Gemini Ultra, the flagship model, would launch to consumers in early 2024 after a round of private tests. While the company didn’t say so explicitly, the understanding at the time was already that the Ultra model would be part of a paid plan, dubbed Bard Advanced at the time and now renamed to Gemini Advanced.

“Gemini Ultra 1.0 is a model that sets the state of the art across a wide range of benchmarks across text, image, audio and video,” Google’s Sissie Hsiao said in a press conference ahead of today’s announcement. “For Google, Gemini is more than just the models. It’s really a shift in how we think about the state-of-the-art technology and the entire ecosystem that we’re building on it, from products that affect billions of users, to the APIs and platforms that developers and businesses use to innovate.”

Hsiao also noted that to reflect that its most advanced technology is at the core of Bard, Google decided to rename it to Gemini (though If Google, with its enormous branding expertise, ever decides to launch a more capable model and wants to call it Aries instead of Gemini Ultra Super Pro Max 3.0, who knows what it’ll call Bard then). And while Google touts Bard’s scores with users, many early users surely never returned after it provided middling results, making a rebrand almost a necessity.

Bard had already switched to Gemini Pro, so for free users, there won’t be any major changes here. Those who opt to pay for Gemini Advanced, though, will get access to the Gemini Ultra 1.0  model. As for how good Gemini Ultra 1.0 really is, we’ll have to try it out ourselves. Google itself was rather vague about its capabilities during this week’s press conference.

“With access to our Ultra 1.0 model, Gemini Advanced is far more capable at highly complex tasks with a range of applications like coding, logical reasoning, following nuanced instructions and creative collaboration, amongst many others,” Hsiao explained. “Not only does Gemini Advanced allow you to write longer prompts, it can also better understand the context of your previous prompts.” Gemini Ultra 1.0, it is worth noting, is also multi-model, which allows you to engage in conversations about images, for example.

Gemini Advanced

As for Gemini Advanced, which is currently the only way to get access to Ultra 1.0, users will have to sign up for the new $20 Google One AI Premium plan. This plan includes all of the benefits of the existing Google One Premium plan (2TB of storage, exclusive Google Photos editing features, Google Meet premium video features and Google Calendar’s enhanced appointment scheduling). In addition, they will get access to the more advanced model wherever Gemini is available. Soon, they will also get access to Gemini in Google Workspace, which will likely be a Microsoft Copilot-like experience, though Google didn’t want to provide any details about this yet.

Indeed, $20 per month is currently the going price for most advanced AI chat tools, including ChatGPT. Google has an advantage here in that it can layer these additional features on top of it without incurring too much of an extra cost (the margins on these existing Google One storage plans are probably quite sizable already, after all).

One nice bonus here: Even if you’re already a Google One subscriber, you’ll also get the free two-month trial if you upgrade to the new AI Premium plan.

https://techcrunch.com/2024/01/07/what-is-google-gemini-ai/