Amazon Bedrock console overview - Amazon Bedrock

Amazon Bedrock console overview

The Amazon Bedrock console provides the following features.

To open the Amazon Bedrock console, sign in at https://console.aws.amazon.com/bedrock/home.

Getting started

From Getting started in the navigation pane, you can get an Overview of the foundation models, examples, and playgrounds that Amazon Bedrock provides. You can also get Examples of the prompts you can use with Amazon Bedrock models.

The examples page shows example prompts for the available models. You can search the examples and filter the list of examples using one or more of the following attributes:

  • Model

  • Modality (text, image, or embedding)

  • Category

  • Provider

Filter the example prompts by choosing the Search in examples edit box and then selecting the filter that you want to apply to the search. Apply multiple filters by again choosing Search in examples and then selecting another filter.

When you choose an example, the Amazon Bedrock console displays the following information about the example:

  • A description of what the example accomplishes.

  • The model name (and model provider) where the example runs.

  • The example prompt and the expected response.

  • The inference configuration parameter settings for the example.

  • The API request that runs the example.

To run the example, choose Open in playground.

Foundation models

From Foundation models in the navigation pane, you can view the available Base models, and group them by various attributes. You can also filter the model view, search for models, and view information about the model providers.

You can customize a base foundation model to improve the model's performance on specific tasks or teach the model a new domain of knowledge. Choose Custom models under foundation models to create and manage your custom models. Customize a model by creating a model customization job with a training dataset that you provide. For more information, see Custom models.

You can experiment with base models and custom models by using the console playgrounds.

Playgrounds

The console playgrounds are where you can experiment with models before deciding to use them in an application. There are three playgrounds.

Chat playground

The chat playground lets you experiment with the chat models that Amazon Bedrock provides. You can submit a chat to a model and the chat playground shows the response from the model and includes model metrics. Optionally, choose Compare mode to compare the output from up to three models. For more information, see Chat playground.

Text playground

The text playground lets you experiment with the text models that Amazon Bedrock provides. You can submit text to a model and the text playground shows the text that the model generates from the prompt. For more information, see Text playground.

Image playground

The image playground lets you experiment with the image models that Amazon Bedrock provides. You can submit a text prompt to a model and the image playground shows the image that the model generates for the prompt. For more information, see Image playground.

In the console, access the playgrounds by choosing Playgrounds in the navigation pane. For more information, see Playgrounds.

Safeguards

Titan Image Generator G1 automatically puts an invisible watermark on all images created by the model. Watermark detection detects if the image was generated by Titan Image Generator G1. To use watermark detection, choose Overview in the left navigation pane and then Build and Test tab. Go to the Safeguards section and choose View watermark detection. For more information, see Watermark detection.

Orchestration

With Amazon Bedrock, you can enable a Retrieval-Augmented Generation (RAG)ß workflow by using knowledge bases to build contextual applications by using the reasoning capabilities of LLMs. To use a knowledge base, choose Orchestration in the left navigation pane and then Knowledge base. For more information, see Knowledge bases for Amazon Bedrock.

Agents for Amazon Bedrock enables developers to configure an agent to complete actions based on organization data and user input. For example you might create an agent to take actions to fulfill a customer's request. To use an Agent, choose Orchestration in the left navigation pane and then Agent. For more information, see Agents for Amazon Bedrock.

Assessment and deployment

As you use Amazon Bedrock models, you need to to assess their performance and to deploy them into your solutions.

With Model Evaluation, you can evaluate and compare model output, and then choose the one best suited for your applications. Choose Assessment and deployment and then choose Model evaluation.

When you configure Provisioned Throughput for a model, you receive a level of throughput at a fixed cost. To provision throughput, choose Assessment and deployment in the navigation pane and then Provisioned Throughput. For more information, see Provisioned Throughput for Amazon Bedrock.

Model access

To use a model in Amazon Bedrock, you must first request access to the model. On the left navigation pane, choose Model access. For more information, see Model access.

Model invocation logging

You can log model invocation events by choosing Settings in the left navigation pane. For more information, see Model invocation logging.