Amazon Bedrock model IDs - Amazon Bedrock

Amazon Bedrock model IDs

Many Amazon Bedrock API operations require the use of a model ID. Refer to the following table to determine where to find the model ID that you need to use.

Use case How to find the model ID
Use a base model Look up the ID in the base model IDs chart
Purchase Provisioned Throughput for a base model Look up the ID in the model IDs for Provisioned Throughput chart and use it as the modelId in the CreateProvisionedModelThroughput request.
Purchase Provisioned Throughput for a custom model Use the name of the custom model or its ARN as the modelId in the CreateProvisionedModelThroughput request.
Use a provisioned model After you create a Provisioned Throughput, it returns a provisionedModelArn. This ARN is the model ID.
Use a custom model Purchase Provisioned Throughput for the custom model and use the returned provisionedModelArn as the model ID.

Amazon Bedrock base model IDs (on-demand throughput)

The following is a list of model IDs for the currently available base models. You use a model ID through the API to identify the base model that you want to use with on-demand throughput, such as in a InvokeModel request, or that you want to customize, such as in a CreateModelCustomizationJob request.

Note

You should regularly check the Model lifecycle page for information about model deprecation and update model IDs as necessary. Once a model has reached end-of-life, the model ID no longer works.

Provider Model name Version Model ID
Amazon Titan Text G1 - Express 1.x amazon.titan-text-express-v1
Amazon Titan Text G1 - Lite 1.x amazon.titan-text-lite-v1
Amazon Titan Text Premier 1.x amazon.titan-text-premier-v1:0
Amazon Titan Embeddings G1 - Text 1.x amazon.titan-embed-text-v1
Amazon Titan Embedding Text v2 1.x amazon.titan-embed-text-v2:0
Amazon Titan Multimodal Embeddings G1 1.x amazon.titan-embed-image-v1
Amazon Titan Image Generator G1 1.x amazon.titan-image-generator-v1
Anthropic Claude 2.0 anthropic.claude-v2
Anthropic Claude 2.1 anthropic.claude-v2:1
Anthropic Claude 3 Sonnet 1.0 anthropic.claude-3-sonnet-20240229-v1:0
Anthropic Claude 3 Haiku 1.0 anthropic.claude-3-haiku-20240307-v1:0
Anthropic Claude 3 Opus 1.0 anthropic.claude-3-opus-20240229-v1:0
Anthropic Claude Instant 1.x anthropic.claude-instant-v1
AI21 Labs Jurassic-2 Mid 1.x ai21.j2-mid-v1
AI21 Labs Jurassic-2 Ultra 1.x ai21.j2-ultra-v1
Cohere Command 14.x cohere.command-text-v14
Cohere Command Light 15.x cohere.command-light-text-v14
Cohere Command R 1.x cohere.command-r-v1:0
Cohere Command R+ 1.x cohere.command-r-plus-v1:0
Cohere Embed English 3.x cohere.embed-english-v3
Cohere Embed Multilingual 3.x cohere.embed-multilingual-v3
Meta Llama 2 Chat 13B 1.x meta.llama2-13b-chat-v1
Meta Llama 2 Chat 70B 1.x meta.llama2-70b-chat-v1
Meta Llama 3 8b Instruct 1.x meta.llama3-8b-instruct-v1:0
Meta Llama 3 70b Instruct 1.x meta.llama3-70b-instruct-v1:0
Mistral AI Mistral 7B Instruct 0.x mistral.mistral-7b-instruct-v0:2
Mistral AI Mixtral 8X7B Instruct 0.x mistral.mixtral-8x7b-instruct-v0:1
Mistral AI Mistral Large 1.x mistral.mistral-large-2402-v1:0
Mistral AI Mistral Small 1.x mistral.mistral-small-2402-v1:0
Stability AI Stable Diffusion XL 0.x stability.stable-diffusion-xl-v0
Stability AI Stable Diffusion XL 1.x stability.stable-diffusion-xl-v1

Amazon Bedrock base model IDs for purchasing Provisioned Throughput

To purchase Provisioned Throughput through the API, use the corresponding model ID when provisioning the model with a CreateProvisionedModelThroughput request. Provisioned Throughput is available for the following models:

Note

Some models have multiple contextual versions whose availability differs by region. For more information, see Model support by AWS Region.

Model name No-commitment purchase supported for base model Model ID for Provisioned Throughput
Amazon Titan Text G1 - Express Yes amazon.titan-text-express-v1:0:8k
Amazon Titan Text G1 - Lite Yes amazon.titan-text-lite-v1:0:4k
Amazon Titan Text Premier (preview) Yes amazon.titan-text-premier-v1:0:32K
Amazon Titan Embeddings G1 - Text Yes amazon.titan-embed-text-v1:2:8k
Amazon Titan Embeddings G1 - Text v2 Yes amazon.titan-embed-text-v2:0:8k
Amazon Titan Multimodal Embeddings G1 Yes amazon.titan-embed-image-v1:0
Amazon Titan Image Generator G1 No amazon.titan-image-generator-v1:0
Anthropic Claude v2 18K Yes anthropic.claude-v2:0:18k
Anthropic Claude v2 100K Yes anthropic.claude-v2:0:100k
Anthropic Claude v2.1 18K Yes anthropic.claude-v2:1:18k
Anthropic Claude v2.1 200K Yes anthropic.claude-v2:1:200k
Anthropic Claude 3 Sonnet 28K Yes anthropic.claude-3-sonnet-20240229-v1:0:28k
Anthropic Claude 3 Sonnet 200K Yes anthropic.claude-3-sonnet-20240229-v1:0:200k
Anthropic Claude 3 Haiku 48K Yes anthropic.claude-3-haiku-20240307-v1:0:48k
Anthropic Claude 3 Haiku 200K Yes anthropic.claude-3-haiku-20240307-v1:0:200k
Anthropic Claude Instant v1 100K Yes anthropic.claude-instant-v1:2:100k
AI21 Labs Jurassic-2 Ultra Yes ai21.j2-ultra-v1:0:8k
Cohere Command Yes cohere.command-text-v14:7:4k
Cohere Command Light Yes cohere.command-light-text-v14:7:4k
Cohere Embed English Yes cohere.embed-english-v3:0:512
Cohere Embed Multilingual Yes cohere.embed-multilingual-v3:0:512
Stable Diffusion XL 1.0 No stability.stable-diffusion-xl-v1:0
Meta Llama 2 Chat 13B No meta.llama2-13b-chat-v1:0:4k
Meta Llama 2 13B No (see note below)
Meta Llama 2 70B No (see note below)
Note

The Meta Llama 2 (non-chat) models can only be used after being customized and after purchasing Provisioned Throughput for them.

The CreateProvisionedModelThroughput response returns a provisionedModelArn. You can use this ARN or the name of the provisioned model in supported Amazon Bedrock operations. For more information about Provisioned Throughput, see Provisioned Throughput for Amazon Bedrock.