Is it possible to use an outside embedding model instead of the ones provided in Amazon Bedrock?

Hi Zach, You have several options available within Amazon Bedrock for embedding model inference. If you’d like to continue using on-demand inference, you can alos consider Cohere Embedding Models (Embed English and Embed Multilingual), or Amazon’s own embedding models such as Titan Text Embeddings V2, Titan Multimodal Embeddings G1, and Titan Embeddings G1 - Text.

Additionally, you can explore other models available in the Amazon Bedrock Marketplace. Once you find a model that fits your needs, you can subscribe to it and deploy it to an endpoint managed by SageMaker AI.

If you're interested in using a different model, you can also easily deploy one using SageMaker JumpStart—a hub of pre-trained, open-source models for a variety of tasks, including embeddings.

Of course, you always have the option to bring your own model and deploy it in SageMaker AI. For example, this article demonstrates how to deploy a Hugging Face embedding model in SageMaker and consume it using LlamaIndex.