Google releases Gemini 1.5 Pro public preview, adding new features

Google on Tuesday launched Gemini 1.5 Pro, an artificial intelligence (AI) model with the largest contextual window, in public preview. The tech giant first announced the AI ​​model in February, and it will be available for developers to try out in Google AI Studio over the next two months. Now, it is open to users for trial use. Enthusiasts can also create or access API keys to build with Large Language Models (LLMs). While being made available to the public, the tech giant has also included several new features in Gemini 1.5 Pro.

The artificial intelligence model is introduce Previewed publicly during the company’s annual Google Cloud Next event. Gemini 1.5 Pro Standard Edition comes with 1,28,000 token context window. In comparison, Gemini 1.0’s context window has 32,000 tokens. There is also a special variant of this model with a huge context window containing a million tokens. Tokens are the primary units of data and can be understood as syllables, words, or subparts of words. The contextual window is the amount of information that the AI ​​model can access to find relevant information based on the keywords in the prompt.

To put this into context, a context window of one million tokens might contain about 7,00,000 words, which is similar to ten average-sized books of 300 pages. This dissemination of information enables AI to understand the broader context and give answers that are more relevant to the user. Additionally, this feature is particularly useful when users want AI to analyze large files to find specific information.

X (formerly Twitter) user Rowan Cheung had early access to the Gemini AI model and posted his findings using the model.in a postal“He said, “I uploaded the entire NBA Dunk Contest last night and asked which dunk scored the highest. Gemini 1.5 is very capable of finding specific perfect 50 dunks and details from its long contextual video understanding! “

The AI ​​model also has some new features. Google has added native audio or voice support, and the Gemini 1.5 Pro can understand spoken prompts. Additionally, a file API for working with files, system directives, and JSON schemas has been added to give developers greater control over their models. It also has multi-mode capabilities and can analyze images and videos. The AI ​​model is currently available in more than 180 countries, including India.