Google Cloud has announced several AI updates, including new models in Vertex AI, enhancements to the Gemini API, and the addition of new languages to Google Translate enabled by AI. Developers can now access the 2 million token context window on Gemini 1.5 Pro without waiting on a waitlist. Users can also apply to be one of the first to use Google’s latest image generator, Image 3, which can create realistic images for various presentations.
New or improved AI models available on the Vertex AI platform include Gemini 1.5 Flash, Gemini 1.5 Pro, and Image 3 in preview. Gemini 1.5 Flash is designed to make generative AI more scalable for high-volume tasks while maintaining quality and context window, as stated by JC Escalante from Ipsos.
Vertex AI will soon offer the light version of Gemini Gemma 2 and Claude 3.5 Sonnet. Context caching and Provisioned Throughput features are also available, with grounding for increased accuracy and high-fidelity grounding mode in experimental preview.
Developers can now run code on Gemini 1.5 Pro and Gemini 1.5 Flash, enabling Python integration within the model for experimenting with generative AI. The Gemini API has been updated to allow for the use of the full 2 million token window on Gemini 1.5 Pro, context caching, and experimentation with Gemma 2 in Google AI Studio.
Google Translate has added Cantonese and 109 other languages, thanks to the PaLM 2 language model. This expansion is the largest ever for the service, with Cantonese being a particularly challenging language to add due to its overlap with Mandarin in writing. PaLM 2 has enabled Google to efficiently aggregate similar languages for inclusion in Google Translate.
Overall, Google Cloud’s latest AI advancements offer developers and users a range of new capabilities and enhancements across various platforms and services, showcasing the continued innovation in AI technology.
Article Source
https://www.techrepublic.com/article/google-cloud-vertex-ai-upgrades/