For the first time in Google is a new model for giving Gemini

On Friday, Google added a new “embroidery” model for the text, and the inclusion of Gemini, to the Gemini developer’s application programming interface.

Input models such as words and phrases are translated into numerical representations, known as implications, which capture the semantic meaning of the text. Connects are used in a set of applications, such as retrieving and classifying documents, in part because they can reduce costs while improving cumin.

Companies including Amazon, Cohere and OpenAi offer models of inclusion through their own application programming facades. Google has provided models that include before, but the inclusion of Gemini is the first training in the Gemini family in artificial intelligence models.

“He was trained in the Gemini model itself, and this inclusion model inherited this Gemini understanding of language and accurate context, which makes it applicable to a wide range of uses,” Google, Google He said in a blog post. “We have trained our model to be noticeably general, as it performed exceptional performance across the various fields, including financing, science, legal, research and more.”

Google claims that the inclusion of Gemini exceeds the performance of the previous gradual inclusion model, and achieves texts-and fulfills competitive performance on the standards of popular inclusion. Compared to Embedding -004, Gemini can also accept more cuts of text and symbol at one time, and supports twice the number of languages ​​(more than 100).

Google indicates that the inclusion of Gemini in a “experimental stage” has a limited capacity and is subject to change. The company wrote in its blog publication: “(w) works for a stable and general issuance in the coming months.”