 |
Today we’re introducing the Amazon Bedrock Marketplace, a new feature that gives you access to over 100 popular, emerging, and specialty foundation models (FMs) through Amazon Bedrock. With this launch, you can now discover, test and deploy new models from enterprise providers like IBM and Nvidia, specialized models like Upstages’ Solar Pro for Korean language processing and Evolutionary Scale’s ESM3 for protein research, along with Amazon’s generic models Bedrock. purpose-built FM from providers such as Anthropic and Meta.
Models deployed using the Amazon Bedrock Marketplace can be accessed through the same standard APIs as serverless models, and for models that are compatible with the Converse API, they can be used with tools such as Amazon Bedrock Agents and the Amazon Bedrock Knowledge Base.
As generative AI continues to reshape the way organizations operate, there is a growing need for specialized models optimized for specific domains, languages or tasks. However, finding and evaluating these models can be challenging and expensive. You have to discover them in different services, create abstractions for using them in your applications, and create complex layers of security and management. Amazon Bedrock Marketplace addresses these challenges by providing a single interface to access both niche and general purpose FM.
Using the Amazon Bedrock Marketplace
To start, in the Amazon Bedrock console, I choose Catalog of models in Foundation models part of the navigation panel. Here I can search for models to help me with a specific use case or language. Search results include both serverless models and models available on the Amazon Bedrock Marketplace. I can filter the results by provider, modality (such as text, image, or audio), or task (such as classifying or summarizing text).
The catalog features models from organizations such as Arcee AI, which creates context-sensitive small language models (SLMs), and Widn.AI, which provides multilingual models.
For example, I’m interested in the IBM Granite models and I’m looking for z models IBM Data and AI.
i choose Granite 3.0 2B instructionslanguage model designed for enterprise applications. Selecting a model will open the model details page where I can see additional information from the model provider, such as the most important information about the model, pricing and usage, including sample API calls.
This particular model requires a subscription and I choose View subscription options.
I’ll review the pricing and legal notices in the subscription dialog. IN Price detailsI see the software price set by the provider. With this model, there are no additional costs on top of the deployed infrastructure. Amazon SageMaker infrastructure costs are billed separately and can be viewed in Amazon SageMaker Pricing.
I choose to continue with this model Subscribe.
Once the subscription is complete, which usually takes a few minutes, I can deploy the model. For Deployment detailsI’m using the default settings and the recommended instance type.
I expand optional Advanced settings. Here I can choose to deploy in a virtual private cloud (VPC) or specify the AWS Identity and Access Management (IAM) role that the deployment uses. Amazon Bedrock Marketplace automatically creates a service role to access the Amazon Simple Storage Service (Amazon S3) buckets where the model weights are stored, but I can choose to use an existing role.
I’ll leave the defaults and finish the deployment.
After a few minutes it is deployed On duty and can be checked in Market deployment page from the navigation bar.
There I can select an endpoint to view details and edit configuration such as the number of instances. For deployment testing I choose Open on the field and ask for some poetry.
I can also choose a model from Chat/text page Playground using a new Marketplace category where the deployed endpoints are listed.
Similarly, I can use the model with other tools such as Amazon Bedrock Agents, Amazon Bedrock Knowledge Base, Amazon Bedrock Prompt Management, Amazon Bedrock Guardrails, and model evaluation by selecting Select Model and selection Marketplace endpoint of the model.
The model I used here is text-to-text, but I can use the Amazon Bedrock Marketplace to deploy models with different modalities. For example, after I deploy Stability AI Stable Diffusion 3.5 Large, I can run a quick test in Amazon Bedrock Picture of the playground.
The models I deployed are now available through the Amazon Bedrock InvokeModel API. When the model is deployed, I can use it with the AWS Command Line Interface (AWS CLI) and any AWS SDKs using the Amazon Resource Name (ARN) endpoint name as the model ID.
For chat-tuned text-to-text models, I can also use the Amazon Bedrock Converse API, which abstracts the differences between models and allows switching between models with a single parameter change.
Things you should know
Amazon Bedrock Marketplace is available in the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Mumbai), Asia Pacific (Seoul), Asia Pacific (Singapore) , Asia Pacific (Sydney), Asia Pacific (Tokyo), Canada (Central), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Paris) and South America (São Paulo).
With Amazon Bedrock Marketplace, you pay a software fee to the third-party model provider (which can be zero, as in the previous example) and a hosting fee based on the type and number of instances you choose for your model endpoints.
Start browsing new models using the model catalog in the Amazon Bedrock console, visit the Amazon Bedrock Marketplace documentation, and submit feedback to AWS re:Post for Amazon Bedrock. On community.aws, you can find detailed technical content and see how our Builder communities use Amazon Bedrock.
— Danilo
