Google LLC’s cloud unit today made several of the generative artificial intelligence tools in its Vertex AI product suite generally available.

The move comes a month after the company first previewed the tools at its Google I/O conference.

Vertex AI is a suite of cloud services that companies can use to build machine learning models, as well as perform related development tasks. The platform includes features for managing neural networks’ training datasets. Once training is complete and a model is deployed, developers can use Vertex AI to monitor it for technical issues.

The first addition to Vertex AI that became generally available today is a service called Model Garden. It provides access to more than 60 generative AI models hosted on Google Cloud. They can be used to generate text, debug code, create images and perform other tasks.

Some of the neural networks available through Model Garden were developed by Google. Others are hosted versions of free AI systems from the open-source ecosystem. According to Google, a subset of the models can be customized by customers for specific tasks.

One of the main highlights of the Model Garden neural network catalog is the Vertex AI PaLM application programming interface. It’s a text generation service based on PaLM 2, a large language model Google debuted earlier this year. It can generate text in multiple languages, as well as perform reasoning tasks such as solving riddles.

PaLM 2 is an enhanced version of a 540 billion-parameter model Google first detailed in 2022. Compared with its predecessor, PaLM 2 generates responses faster. Additionally, it’s more adept at performing tasks such as translation that involve processing text written in multiple languages.

Model Garden also offers access to Codey, an AI system that became available in public preview this morning. It’s a large language model specifically optimized for programming tasks.

According to Google, Codey can analyze a snippet of code written by a developer and suggest the next few lines that should be added to the file. Moreover, it includes a chatbot interface that supports natural language queries. Developers can ask it to generate code, debug existing files and explain programming concepts.

Google Cloud is rolling out Model Garden alongside a tool called Generative AI Studio that also became generally available today. The tool is aimed at making it easier for developers to use the search giant’s neural networks.

Using Generative AI Studio, developers can customize the weights of some Google Cloud neural networks. Weights are the components of a model that determine which data points it takes into account when making a decision. Customizing them makes it possible to improve the quality of the model’s output.

According to Google, developers can also use Generative AI Studio to test what answers a model generates in response to specific prompts. Such testing helps software teams determine which of the search giant’s models is most suitable for a given application project.

“Model Garden and Generative AI Studio leverage Google Cloud’s tight partnership with Google Research and Google DeepMind, making it easy for developers and data scientists to use, customize, and deploy models,” June Yang, Google Cloud’s vice president of cloud AI and industry solutions, detailed in a blog post.

Vertex AI’s generative AI tools have already been adopted by a number of early customers. Google Cloud disclosed today that DataStax Inc., Canva Inc. and GitLab Inc. are among the companies that are using the tools to enhance their software products.

Image: Google