Skip to main content

AWS Bedrock

Amazon Bedrock is a fully managed service that offers a choice of high-performing foundation models (FMs) from leading AI companies like AI21 Labs, Anthropic, Cohere, Meta, Stability AI, and Amazon via a single API, along with a broad set of capabilities you need to build generative AI applications with security, privacy, and responsible AI.

Requisites​

To use AWS Bedrock models, you need to have the following credential information.

Required credentials:

  • AWS_ACCESS_KEY_ID: Your AWS Bedrock Access Key ID. It can be found on your security credentials page.
  • AWS_SECRET_ACCESS_KEY: Your AWS Bedrock Secret Access Key. It can be found on your security credentials page.
  • AWS_REGION: The AWS region where your AWS Bedrock model is hosted.

Supported Models​

All models provided by AWS Bedrock has the following properties:

  • Function call: Not supported
  • Streaming: Not supported

All models accept the following extra configs when generating texts:

  • temperature
  • top_p
  • top_k
  • max_tokens
  • stop

Jurassic-2 Mid​

Model schema id: aws_bedrock/ai21/j2-mid-v1

A versatile, mid-sized language model from AI21 Labs, designed for efficient, high-quality natural language processing tasks across various domains.

Properties:

  • Input Token Limit: 8191

Jurassic-2 Ultra​

Model schema id: aws_bedrock/ai21/j2-ultra-v1

Jurassic-2 Ultra is AI21 Labs' state-of-the-art language model, offering powerful natural language understanding and generation for a wide range of applications and industries.

Properties:

  • Input Token Limit: 8191

Titan Text Express v1​

Model schema id: aws_bedrock/amazon/titan-text-express-v1

Titan Text Express v1 is a robust text generation AI, designed for efficient, high-quality content creation across diverse applications with speed and accuracy.

Properties:

  • Input Token Limit: 8192

Titan Text Lite v1​

Model schema id: aws_bedrock/amazon/titan-text-lite-v1

Titan Text Lite v1 offers a streamlined, cost-effective solution for text generation, ideal for lightweight applications needing fast, reliable AI-generated content.

Properties:

  • Input Token Limit: 4096

Claude Instant 1.2​

Model schema id: aws_bedrock/anthropic/claude-instant-v1

Claude Instant 1.2 is a smaller, faster, and more efficient version of Claude 2.1, designed for real-time applications.

Properties:

  • Input Token Limit: 100000

Claude-v2​

Model schema id: aws_bedrock/anthropic/claude-v2

Claude 2 enhanced user interaction, improved in coding, math, logical reasoning, handled 100K tokens, and focused on safety and performance.

Properties:

  • Input Token Limit: 100000

Claude-v2.1​

Model schema id: aws_bedrock/anthropic/claude-v2.1

Claude 2.1 increased accuracy, doubled context window, introduced experimental features, and emphasized honesty and detailed engagement.

Properties:

  • Input Token Limit: 200000

Command Light Text v14​

Model schema id: aws_bedrock/cohere/command-light-text-v14

Command Light Text v14, by Cohere, offers a lightweight, efficient AI model for text processing, ideal for resource-constrained environments and applications.

Properties:

  • Input Token Limit: 4000

Command Text v14​

Model schema id: aws_bedrock/cohere/command-text-v14

Cohere's Command Text v14 is a robust AI model designed for complex text processing, enabling advanced understanding and generation for demanding applications.

Properties:

  • Input Token Limit: 4000

Llama2 13B Chat v1​

Model schema id: aws_bedrock/meta/llama2-13b-chat-v1

A versatile AI chat model by Meta, designed for efficient and effective conversational understanding and generation across diverse topics.

Properties:

  • Input Token Limit: 4096

Llama2 70B Chat v1​

Model schema id: aws_bedrock/meta/llama2-70b-chat-v1

An advanced version of Meta's AI chat model, offering deeper insights and more nuanced conversation capabilities with its 70 billion parameter design.

Properties:

  • Input Token Limit: 4096