AWS launches Bedrock Marketplace for AI models

AWS has launched Amazon Bedrock Marketplace – a hub for foundation models from the likes of Nvidia, IBM, and over 100 more.

Announced during AWS’ re:Invent conference in Las Vegas, the marketplace aims to make it easy to discover and deploy both general-purpose and specialised AI models.

The idea is to have models from both large enterprises and startups available through a single interface.

The Bedrock Marketplace allows users to test and deploy models with consistent APIs and tools such as Amazon Bedrock Agents, Knowledge Bases and Guardrails.

“Zendesk has an opportunity to access specialised models… from Amazon Bedrock Marketplace’s vast catalogue and incorporate them into their generative AI features for translation use cases, further personalising and localising customer service requests,” Jon Aniano, senior vice president of product at Zendesk said.

More AWS announcements today

This announcement comes alongside updates to SageMaker HyperPod, AWS’s managed platform for building and fine-tuning large language models (LLMs). New features include flexible training plans, which optimise resource allocation for model training, and HyperPod Recipes, designed to streamline the fine-tuning of open-weight models like Meta’s Llama. 

According to AWS, enterprises can also pool GPU resources for dynamic allocation, improving cost efficiency by up to 40%.

AWS also announced that SageMaker – its machine learning platform – can now support third-party apps. This means businesses can use popular AI tools directly within SageMaker without worrying about data leaving the platform. 

According to AWS, this allows users to manage their machine learning projects securely in one place without having to find complex workarounds.

AWS’s announcements follow the debut of its Amazon Nova multimodal foundation models yesterday. The company also announced multi-agents and enhanced guardrails in Amazon Bedrock as well as updates to Amazon Q Developer and Q Business, and the general availability of its Trainium 2 chips.

The author travelled to Re:Invent in Las Vegas as a guest of AWS.

Never miss a story: sign up to SmartCompany’s free daily newsletter and find our best stories on LinkedIn.