Google's first true ChatGPT competitor, Gemini, has been released onto the world stage. The advanced large language model is being touted as a new and improved version of the lackluster AI chatbot Bard, which is now exiting stage left.
“We’re entering the next chapter of our Gemini era,” Alphabet and Google CEO Sundar Pichai said Thursday about the company’s largest, most flexible, and most capable AI model to date.
The chatbot’s first and most powerful iteration, Gemini Ultra 1.0, is now being offered to try on the Play store, complete with its own brand new app, and in the iOS Apple store under the Google app.
Curious users can also test out Gemini by visiting its new dedicated Google web address (gemini.google.com)
Now released in 150 countries, the powerful multi-modal large language model (LLM) comes in three versions according to Google.
- Gemini Ultra - largest model for highly complex tasks.
- Gemini Pro - best model for scaling across a wide range of tasks.
- Gemini Nano - most efficient model for on-device tasks.
“Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding), which uses a combination of 57 subjects such as math, physics, history, law, medicine, and ethics for testing both world knowledge and problem-solving abilities,” said Demis Hassabis, CEO and Co-Founder of Google DeepMind.
On those benchmark subjects, 90% of the time, Gemini either outperformed its rival ChatGPT in “sophisticated reasoning capabilities” or was fairly similar percentage-wise when returning text, images, video, audio, and even advanced coding.
Users will be able to subscribe to Gemini Ultra for $19.99 per month, with the first two months free, driving up the competition with ChatGPT Pro and Microsoft, who both offer similar advanced model subscriptions for the same price per month.
The Gemini ‘Google One AI Premium plan’ will also be integrated with Gmail, Google's productivity suite, and include two terabytes of cloud storage, which is normally $9.99 per month.
Similar to ChatGPT, Gemini works using a prompt and response model with the LLM answering a user’s query based on what the chatbot already knows and by scouring Google services.
When it comes to user risk, Hassabis states that Gemini “has the most comprehensive safety evaluations of any Google AI model to date, including for bias and toxicity.”
As of Thursday, users with Android devices can choose to opt into Gemini as the default assistant and access the chatbot through the app, the power button, or by saying "Hey Google."
A Gemini smartphone will also start to roll out access in the US, Latin America, and more. Although currently only available in English, support for Japanese and Korean languages will also be released next week.
Last February, Google lost more a $100 billion in market value after millions of viewers watched Bard give inaccurate answers during a hyped-up live-streamed press event introducing the AI chatbot for the first time to the public.
And just in case there is any confusion, Google wants to remind users that its Gemini chatbot is entirely AI generated.
Even though “Gemini might sound like a human… it doesn’t have its own thoughts or feelings,” Google stated.
More from Cybernews:
Subscribe to our newsletter