Google announced today its Cloud portfolio expansion with the availability of the Vertex AI platform to its enterprise customers. Google focused firmly on its AI venture at Google I/O this year; it is trying to ensure the widest availability for consumers and enterprises to take advantage of Generative AI.
Google opened the doors to Generative AI support on Vertex AI on June 2023, with baked-in security, data governance, and scalability across the board. This brings access to powerful foundation models and customizing tools to customers to meet their needs. Following suit, today, Google announced the general availability (GA) of four foundation models for Vertex AI:
- Imagen
- PaLM 2 for Chat
- Codey
- Chirp
It made APIs of these models accessible through a new tool in its Cloud Platform named Model Garden. The garden has access to 30 languages, 58 visions, and many more foundation models.
These models can be used to do prompt design and tuning on Generative AI Studio, which has step-by-step guides and instructions on how to take better advantage of the available tools for customers.
Google listed the key features of the foundation models as follows:
- Imagen includes four key features:
- Image generation for creating studio-grade images at scale
- Image editing to edit generated or existing images via text prompts
- Image captioning for creating captions of images at scale
- Visual Question & Answering (VQA) for interacting with, analyzing, and explaining images
- PaLM 2 for Chat follows the general availability of PaLM 2 for Text in June
Codey supports code generation, completion, and code chat
Chirp supports multilingual Speech AI
The blog post also mentions Multimodal Embeddings API as a preview. This will enable customers to take advantage of the powerful Vertex AI using proprietary data, including images, texts, interchangeable vectors, and more. This capability will come with a privacy-preserving approach for customers.