Dive Brief:
- AWS expanded its generative AI offerings, augmenting its foundation model API service Bedrock with Anthropic’s Claude 2, Stability AI’s Stable Diffusion XL 1.0 and two Cohere models, the company announced Wednesday.
- The four additions join the multimodel roster of generative AI technologies available via Bedrock since its April launch. Customers access the different models through a secure API and can use their data to customize specific functions.
- AWS is molding Bedrock into a nascent marketplace for generative AI tools and technologies. “No one model will rule them all,” Swami Sivasubramanian, VP of database, analytics and ML at AWS, said during Wednesday’s AWS Summit New York. “Rather, organizations will need to be able to choose the right model for the right job.”
Dive Insight:
Convenient access to a broad array of enterprise technologies, including “as a Service” infrastructure, platforms and software, has long been part of cloud’s appeal.
Generative AI joined the catalog of offerings earlier this year, as the three-largest hyperscalers — AWS, Microsoft and Google Cloud — began jockeying for dominance of the emerging tech category.
While Amazon’s general purpose Titan foundation models are available on Bedrock, AWS is setting up a marketplace for foundation models that fulfill a wide range of business needs.
Claude 2, the second iteration of AI startup Anthropic’s LLM technology, is similar to OpenAI’s ChatGPT tools in function. The new version can ingest hundreds of pages of text and produce longer documents than its predecessor, according to AWS.
Stability AI’s SDXL 1.0 suite of models specialize in generating images and video from text prompts. Cohere offers two separate models, Command and Embed, for generating and understanding text, the announcement said.
In addition to increasing the number of models available on Bedrock, AWS introduced a low-code automation “agents” tool, and brought online the NVIDIA H100 Tensor Core GPU-powered compute capabilities promised in March.
Two additional tools are now in preview, AWS said Wednesday. An ML-augmented vector search tool for OpenSearch Serverless is accessible to users on a limited basis for testing, as are generative AI enhancements to AWS’s QuickSight BI service, the company said in separate announcements.