Image: Google
Google today announced major updates to Vertex AI, the tech giant’s suite of AI models and technologies for enterprise customers.
“Today’s news is aimed at helping organizations build transformative experiences with enterprise-ready generative AI,” a spokesperson for Google told iPhone in Canada. The newly announced improvements focus largely on Google’s Gemini 1.5 Flash, Gemini 1.5 Pro, Imagen 3, and Gemma 2 AI models.
Gemini 1.5 Flash was unveiled last month and was previously only available in preview. The AI model is a significant upgrade over its predecessor, Gemini 1.0 Ultra, and is now generally available to all Vertex AI users.
“Gemini 1.5 Flash combines low latency, competitive pricing, and our groundbreaking 1 million-token context window, making it an excellent option for a wide variety of use cases at scale, from retail chat agents, to document processing, to research agents that can synthesize entire repositories,” said Google.
“Tokens” are the smallest units of data an AI model needs to process to respond to a prompt. Token windows represent how much data a model can process without losing contextual track of all the prompts and responses exchanged within the session.
Gemini 1.5 Pro, meanwhile, was announced earlier this year and is now publicly available with its massive 2-million token context window. According to Google, “Gemini 1.5 Pro is equipped to unlock unique multimodal use cases that no other model can handle.”
Both Gemini 1.5 Pro and Flash now offer context caching to provide lower cost and higher speed.
Imagen 3, a new iteration of Google’s image generation model, is now available in preview for Vertex AI customers with early access. The generative AI model is over 40% faster than its predecessor, Imagen 2, and features better prompt understanding and instruction-following, photo-realistic generations, greater control over text rendering, multi-language support, and safety features like digital watermarking.
Google also announced that Gemma 2, a lightweight, cutting-edge open model that comes in both 9-billion (9B) and 27-billion (27B) parameter sizes, will be available to researchers and developers across the globe as part of Vertex AI starting next month. “Gemma 2 is much more powerful and efficient than the first generation, with significant safety advancements built in,” the company said.
Furthermore, Google is kicking off the general availability of Grounding with Google Search, a set of measures to reduce model hallucinations for Vertex AI customers. With this feature, Gemini model outputs can be backed up with data from Google Search to reduce inaccuracies and ensure fresh, high-quality information.
In addition, starting next quarter, Vertex AI will offer a new service allowing customers to ground their AI agents with specialized third-party data from partners such as Moody’s, MSCI, Thomson Reuters, and Zoominfo. There will also be a “Grounding with High-Fidelity” mode that will generate responses exclusively from the provided context instead of relying on a model’s built-in knowledge base.
In its announcement, Google included testimonials from existing Vertex AI customers — UberEats, Ipsos, Jasper, Shutterstock, Quora, and more — who are using the models and tools provided by Vertex AI to create their own AI agents and many other applications.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
