Google’s pre-generative large language model Gemini is growing up a bit with a .5 upgrade rolling out to users in over 180 countries now.
Google Gemini 1.5 Pro is available now through Google AI Studio for developers and introduces new audio and video modalities, API improvements, and a new embedded model with additional performance optimizations.
According to the Google for Developers blog the new Audio and Video Modalities include “audio (speech) understanding in both the Gemini API and Google AI Studio.”
Developers can also tap into the ability for Gemini 1.5 Pro’s support for reasoning across images (frames) and audio (speech) when videos are uploaded into the studio soon.
As for the API improvements, there are System Instructions that help to define roles, formats, goals, and rules for AI behavior in different instances. There is a JSON Mode that helps instruct models to output into JSON objects. Google also notes there are improvements to function calling so developers can assign text, function calls, or simply the function altogether.
Lastly, there is a new text embedding model present in Gemini 1.5 Pro thanks to API support that aids in stronger retrival performance, while also outperforming the previous models, according to Google.
Google is promising more Gemini API and AI Studio improvements over the coming weeks.
With OpenAI set to debut its GPT-5 model soon, we should see how Google’s latest model improvements stack up against OpenAI and subsequently Microsoft, soon.
