Google’s answer to GPT 4 may become reality very soon in the form of Bard Advanced, a new paid tier for Google’s Bard chatbot powered by Gemini Ultra.

Bard Advanced was originally announced during Google’s Gemini launch in late December, but no details were revealed back then. However, we now have evidence that Google is actively working on Bard Advanced thanks to new information spotted on Google’s web code. This evidence was spotted by X users Bedros Pamboukian and Dylan Roussel.

The X thread below shows a list of features Google is currently working on.

The evidence suggests that Bard Advanced will be available through a paid subscription via the Google One program, which also grants users extra Google Photos storage and other perks. The first three months of Bard Advanced will reportedly be free upon subscription.

Similar to OpenAI’s ChatGPT, a paid subscription for Bard will also unlock the ability to create custom chatbots as well as a dedicated chatbot store. This feature is codenamed “Mokoto” while the chatbot store is internally known as “Sparkle” at the moment. It remains to be seen when Mokoto and Sparkle will become official or how they will be priced.

Google’s introduction of a premium subscription model could be seen as a strategic move to offset the substantial expenses involved in operating large language models, particularly the highly advanced ones like Ultra.

Gemini Ultra

Google Gemini was launched in early December 2023 in three tiers including Gemini Nano, Gemini Pro, and Gemini Ultra. Gemini Nano, the smallest AI model in the lineup, will be powering consumer devices including smartphones. It is already working behind the scenes on Google’s Pixel 8 Pro.

Gemini Pro is currently the brains behind the Bard chatbot, but users should soon be able to use the more advanced Gemini Ultra through Bard Advanced, which will be capable of more sophisticated tasks such as advanced math and enhanced reasoning.

This would align with OpenAI’s approach for its GPT models deployed in ChatGPT. GPT 3.5, the less capable model is available for free while users have to pay a monthly fee to use GPT 4. This also grants access to DALL-E 3, which is OpenAI’s flagship image generator AI. Perhaps, Google will add image generation features to Bard Advanced as well, but that remains to be seen.

Via: The Decoder