Pricing overview

Amazon Bedrock is a fully managed service that offers a choice of high-performing foundation models (FMs) through a single API, along with a broad set of capabilities you need to build generative AI applications with security, privacy, and responsible AI.

With Amazon Bedrock, you will be charged for:

• Model inference
• Model customization
• You have a choice of two pricing plans for inference:
  • 1. On-Demand and Batch
    Pay-as-you-go pricing without time-based commitments.
  • 2. Provisioned Throughput:
    Provision throughput to meet performance requirements in exchange for a time-based commitment.

Pricing Models

On Demand and Batch

• On-Demand Mode: Pay only for what you use with no term commitments.
  • Text-Generation Models: Charged per input and output token.
  • Embeddings Models: Charged per input token.
  • Image-Generation Models: Charged per image generated.
  • Cross-Region Inference: Supports using compute across different AWS Regions to manage traffic bursts, with no extra charge.
• Batch Mode: Submit prompts as a single input file and receive responses in an output file.
  • Responses are stored in an Amazon S3 bucket for future access.
  • Batch inference pricing is 50% lower than on-demand pricing for select models from providers like Anthropic, Meta, Mistral AI, and Amazon.

Provisioned Throughput

• Provisioned Throughput Mode: Purchase model units for a specific base or custom model.
  • Designed for large, consistent inference workloads needing guaranteed throughput.
  • Custom models are only available with this mode.
  • Model Unit: Provides a defined throughput (tokens processed per minute).
  • Pricing: Charged by the hour with a choice of 1-month or 6-month commitment terms.

Custom Model Import

• Custom Model Import: Import your customized models into Amazon Bedrock to use them like other hosted models.
  • No Charge: Importing a custom model to Bedrock is free.
  • On-Demand Serving: Imported models are available on-demand with no control plane actions required.
  • Inference Pricing: Charged based on the number of model copies needed for inference and their active duration (billed in 5-minute increments).
  • Model Copy Cost: Pricing depends on factors like architecture, context length, AWS Region, compute version, and model size tier.

‍