Google launches Gemini Advanced subscription with its most powerful Ultra model



summary
Summary

After months of anticipation, Google has flipped the big switch that turns Bard into Gemini and gives it access to its most powerful language model.

Google has renamed its AI chatbot Bard to Gemini and introduced Gemini Advanced, a paid subscription. This gives users access to Google’s most powerful language model to date, Gemini Ultra 1.0. The move has been on the cards for a while, so it comes as no big surprise. Even the full text of the official announcement was already known.

Image: Screenshot/THE DECODER

Gemini Advanced is Google’s alternative to OpenAI’s paid membership ChatGPT Plus, which costs $20 per month. Gemini Advanced is slightly more expensive at $21.99. For that price, Google also offers two terabytes of cloud storage in Google Drive and other benefits from the Google One subscription. Gemini Advanced is available for a two-month free trial.

Google Cloud integration coming later

Google’s strength, the integration of the language model into an existing huge ecosystem, will not come into play from the start. Gemini Ultra 1.0 will only be available in Gmail and Docs with the Advanced subscription at a later date. Also announced, but not yet available, are a larger context window, extended multimodal capabilities, better programming features, and file upload and analysis – features that have long since found their way into ChatGPT in recent months. Single images can already be uploaded.

Ad

Ad

In addition to the new model, Google has also redesigned the Gemini interface to reduce visual distractions. Readability and navigation have also been improved. Google notes that although Ultra 1.0 is primarily trained on English material, it can respond in any language in which Gemini is available. However, the smartphone app will initially be available only in English and in the United States.

Google’s answer to Microsoft Copilot

Gemini Ultra was launched in late 2023 as a versatile model for tasks such as coding, reasoning, and creative collaboration. According to Google, it outperforms GPT-4 in 30 out of 32 benchmarks and is the first model to match the performance of a human expert on the MMLU benchmark. Independent confirmation of these claims has been lacking due to limited availability, but this should change soon with the release. Microsoft has, however, already debunked the somewhat embellished benchmark results.

In addition to Bard, the workspace feature previously known as “Duet AI” will also be renamed to Gemini. With Gemini Advanced, Google is trying to reduce Microsoft’s lead in the target group of professional AI users. Microsoft has continued to expand its Copilot platform and integration with the Office suite with regular updates.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top