Another day, another AI chatbot: Google has just released a new AI model that it is calling “Gemini.” It now joins other top AI chatbots like ChatGPT and Microsoft’s Copilot.
The new generative AI can handle images and video as well as text. The initial model is rolling out today with Google’s Bard chatbot.
It’s currently only available for the English language setting of Bard and will continue rolling out to many more Google products in the “coming months.” Here’s what to know about the new technology — which is Google’s “largest and most capable” AI — and how to use it.
What Makes Gemini Stand Out?
The big reason why Google bills Gemini as its most capable model yet is the engine’s multimodal capacity. In other words, the new tech can seamlessly handle generating and reading a wider-than-ever variety of data. Here’s the fast list of all the data types Gemini works with:
- Text
- Code
- Audio
- Image
- Video
This just in! View
the top business tech deals for 2024 👨💻
Gemini is also the most flexible model Google has delivered. The main tech being debuted today is called Gemini 1.0, but it comes in three different “sizes.”
- Gemini Ultra — largest and most capable model for highly complex tasks
- Gemini Pro — best for scaling across a wide range of tasks (Bard has this version)
- Gemini Nano — most efficient model for on-device tasks
Google is a sprawling corporation, so they need the Gemini brand to cover a vast range of uses, from huge data centers to tiny on-device uses like message suggestions. It makes sense that they’d be issuing multiple versions at once in order to deal with all their wildly different business use cases.
What Products Is Gemini Coming to and When?
The initial Bard English language rollout is available now for more than 170 countries and territories. If you want to try it out, head over to Bard and ask it if it’s using Gemini yet.
Also available now: A version of Gemini is on the Pixel 8 OS, guiding users through suggested messages.
Later this month, Gemini will become available to developers: The Google Cloud API will have Gemini starting on December 13th.
As for additional Gemini integrations, you should expect a lot more news in “coming months,” when Google says the product will become available on more products and services like Search, Ads, Chrome and Duet AI.
Related: How To Use Google’s New AI Image Generator in Search
Gemini Ultra Is “Coming Soon” in 2024
Google is “currently completing extensive trust and safety checks” for their most powerful AI model yet, Gemini Ultra. They plan to roll it out sometime in 2024.
What will Ultra add that the current Gemini models don’t have? A more fine-tuned functionality, boosted thanks to the inclusion of reinforcement learning from human feedback (called RLHF). Select developers and experts will gain early access for feedback purposes.
Also coming even sooner in 2024 is Bard Advanced, a Bard interface that packages up the latest models, including Ultra, once it’s available. Until then, it’s easy to start using the Gemini Pro model now: Just head over to Bard and start talking just like you would with any other generative AI.
