Class: Aws::BedrockRuntime::Client

Inherits:

Seahorse::Client::Base

• Object
• Seahorse::Client::Base
• Aws::BedrockRuntime::Client

Includes:

ClientStubs

Defined in:

gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb

Overview

An API client for BedrockRuntime. To construct a client, you need to configure a :region and :credentials.

client = Aws::BedrockRuntime::Client.new(
  region: region_name,
  credentials: credentials,
  # ...
)

For details on configuring region and credentials see the developer guide.

See #initialize for a full list of supported configuration options.

Instance Attribute Summary

Attributes inherited from Seahorse::Client::Base

#config, #handlers

API Operations

• #apply_guardrail(params = {}) ⇒ Types::ApplyGuardrailResponse

  The action to apply a guardrail.

• #converse(params = {}) ⇒ Types::ConverseResponse

  Sends messages to the specified Amazon Bedrock model.

• #converse_stream(params = {}) ⇒ Types::ConverseStreamResponse

  Sends messages to the specified Amazon Bedrock model and returns the response in a stream.

• #get_async_invoke(params = {}) ⇒ Types::GetAsyncInvokeResponse

  Retrieve information about an asynchronous invocation.

• #invoke_model(params = {}) ⇒ Types::InvokeModelResponse

  Invokes the specified Amazon Bedrock model to run inference using the prompt and inference parameters provided in the request body.

• #invoke_model_with_response_stream(params = {}) ⇒ Types::InvokeModelWithResponseStreamResponse

  Invoke the specified Amazon Bedrock model to run inference using the prompt and inference parameters provided in the request body.

• #list_async_invokes(params = {}) ⇒ Types::ListAsyncInvokesResponse

  Lists asynchronous invocations.

• #start_async_invoke(params = {}) ⇒ Types::StartAsyncInvokeResponse

  Starts an asynchronous invocation.

Instance Method Summary

• #initialize(options) ⇒ Client constructor

  A new instance of Client.

Methods included from ClientStubs

#api_requests, #stub_data, #stub_responses

Methods inherited from Seahorse::Client::Base

add_plugin, api, clear_plugins, define, new, #operation_names, plugins, remove_plugin, set_api, set_plugins

Methods included from Seahorse::Client::HandlerBuilder

#handle, #handle_request, #handle_response

Constructor Details

#initialize(options) ⇒ Client

Returns a new instance of Client.

Parameters:

• options (Hash)

Options Hash (options):

• :plugins (Array<Seahorse::Client::Plugin>) — default: []] —

  A list of plugins to apply to the client. Each plugin is either a class name or an instance of a plugin class.

• :credentials (required, Aws::CredentialProvider) —

  Your AWS credentials. This can be an instance of any one of the following classes:

  • Aws::Credentials - Used for configuring static, non-refreshing credentials.

  • Aws::SharedCredentials - Used for loading static credentials from a shared file, such as ~/.aws/config.

  • Aws::AssumeRoleCredentials - Used when you need to assume a role.

  • Aws::AssumeRoleWebIdentityCredentials - Used when you need to assume a role after providing credentials via the web.

  • Aws::SSOCredentials - Used for loading credentials from AWS SSO using an access token generated from aws login.

  • Aws::ProcessCredentials - Used for loading credentials from a process that outputs to stdout.

  • Aws::InstanceProfileCredentials - Used for loading credentials from an EC2 IMDS on an EC2 instance.

  • Aws::ECSCredentials - Used for loading credentials from instances running in ECS.

  • Aws::CognitoIdentityCredentials - Used for loading credentials from the Cognito Identity service.

  When :credentials are not configured directly, the following locations will be searched for credentials:

  • Aws.config[:credentials]
  • The :access_key_id, :secret_access_key, :session_token, and :account_id options.
  • ENV['AWS_ACCESS_KEY_ID'], ENV['AWS_SECRET_ACCESS_KEY'], ENV['AWS_SESSION_TOKEN'], and ENV['AWS_ACCOUNT_ID']
  • ~/.aws/credentials
  • ~/.aws/config
  • EC2/ECS IMDS instance profile - When used by default, the timeouts are very aggressive. Construct and pass an instance of Aws::InstanceProfileCredentials or Aws::ECSCredentials to enable retries and extended timeouts. Instance profile credential fetching can be disabled by setting ENV['AWS_EC2_METADATA_DISABLED'] to true.
• :region (required, String) —

  The AWS region to connect to. The configured :region is used to determine the service :endpoint. When not passed, a default :region is searched for in the following locations:

  • Aws.config[:region]
  • ENV['AWS_REGION']
  • ENV['AMAZON_REGION']
  • ENV['AWS_DEFAULT_REGION']
  • ~/.aws/credentials
  • ~/.aws/config
• :access_key_id (String)
• :account_id (String)
• :active_endpoint_cache (Boolean) — default: false —

  When set to true, a thread polling for endpoints will be running in the background every 60 secs (default). Defaults to false.

• :adaptive_retry_wait_to_fill (Boolean) — default: true —

  Used only in adaptive retry mode. When true, the request will sleep until there is sufficent client side capacity to retry the request. When false, the request will raise a RetryCapacityNotAvailableError and will not retry instead of sleeping.

• :client_side_monitoring (Boolean) — default: false —

  When true, client-side metrics will be collected for all API requests from this client.

• :client_side_monitoring_client_id (String) — default: "" —

  Allows you to provide an identifier for this client which will be attached to all generated client side metrics. Defaults to an empty string.

• :client_side_monitoring_host (String) — default: "127.0.0.1" —

  Allows you to specify the DNS hostname or IPv4 or IPv6 address that the client side monitoring agent is running on, where client metrics will be published via UDP.

• :client_side_monitoring_port (Integer) — default: 31000 —

  Required for publishing client metrics. The port that the client side monitoring agent is running on, where client metrics will be published via UDP.

• :client_side_monitoring_publisher (Aws::ClientSideMonitoring::Publisher) — default: Aws::ClientSideMonitoring::Publisher —

  Allows you to provide a custom client-side monitoring publisher class. By default, will use the Client Side Monitoring Agent Publisher.

• :convert_params (Boolean) — default: true —

  When true, an attempt is made to coerce request parameters into the required types.

• :correct_clock_skew (Boolean) — default: true —

  Used only in standard and adaptive retry modes. Specifies whether to apply a clock skew correction and retry requests with skewed client clocks.

• :defaults_mode (String) — default: "legacy" —

  See DefaultsModeConfiguration for a list of the accepted modes and the configuration defaults that are included.

• :disable_host_prefix_injection (Boolean) — default: false —

  Set to true to disable SDK automatically adding host prefix to default service endpoint when available.

• :disable_request_compression (Boolean) — default: false —

  When set to 'true' the request body will not be compressed for supported operations.

• :endpoint (String, URI::HTTPS, URI::HTTP) —

  Normally you should not configure the :endpoint option directly. This is normally constructed from the :region option. Configuring :endpoint is normally reserved for connecting to test or custom endpoints. The endpoint should be a URI formatted like:

  'http://example.com'
  'https://example.com'
  'http://example.com:123'
  

• :endpoint_cache_max_entries (Integer) — default: 1000 —

  Used for the maximum size limit of the LRU cache storing endpoints data for endpoint discovery enabled operations. Defaults to 1000.

• :endpoint_cache_max_threads (Integer) — default: 10 —

  Used for the maximum threads in use for polling endpoints to be cached, defaults to 10.

• :endpoint_cache_poll_interval (Integer) — default: 60 —

  When :endpoint_discovery and :active_endpoint_cache is enabled, Use this option to config the time interval in seconds for making requests fetching endpoints information. Defaults to 60 sec.

• :endpoint_discovery (Boolean) — default: false —

  When set to true, endpoint discovery will be enabled for operations when available.

• :event_stream_handler (Proc) —

  When an EventStream or Proc object is provided, it will be used as callback for each chunk of event stream response received along the way.

• :ignore_configured_endpoint_urls (Boolean) —

  Setting to true disables use of endpoint URLs provided via environment variables and the shared configuration file.

• :input_event_stream_handler (Proc) —

  When an EventStream or Proc object is provided, it can be used for sending events for the event stream.

• :log_formatter (Aws::Log::Formatter) — default: Aws::Log::Formatter.default —

  The log formatter.

• :log_level (Symbol) — default: :info —

  The log level to send messages to the :logger at.

• :logger (Logger) —

  The Logger instance to send log messages to. If this option is not set, logging will be disabled.

• :max_attempts (Integer) — default: 3 —

  An integer representing the maximum number attempts that will be made for a single request, including the initial attempt. For example, setting this value to 5 will result in a request being retried up to 4 times. Used in standard and adaptive retry modes.

• :output_event_stream_handler (Proc) —

  When an EventStream or Proc object is provided, it will be used as callback for each chunk of event stream response received along the way.

• :profile (String) — default: "default" —

  Used when loading credentials from the shared credentials file at HOME/.aws/credentials. When not specified, 'default' is used.

• :request_min_compression_size_bytes (Integer) — default: 10240 —

  The minimum size in bytes that triggers compression for request bodies. The value must be non-negative integer value between 0 and 10485780 bytes inclusive.

• :retry_backoff (Proc) —

  A proc or lambda used for backoff. Defaults to 2**retries * retry_base_delay. This option is only used in the legacy retry mode.

• :retry_base_delay (Float) — default: 0.3 —

  The base delay in seconds used by the default backoff function. This option is only used in the legacy retry mode.

• :retry_jitter (Symbol) — default: :none —

  A delay randomiser function used by the default backoff function. Some predefined functions can be referenced by name - :none, :equal, :full, otherwise a Proc that takes and returns a number. This option is only used in the legacy retry mode.

  @see https://www.awsarchitectureblog.com/2015/03/backoff.html

• :retry_limit (Integer) — default: 3 —

  The maximum number of times to retry failed requests. Only ~ 500 level server errors and certain ~ 400 level client errors are retried. Generally, these are throttling errors, data checksum errors, networking errors, timeout errors, auth errors, endpoint discovery, and errors from expired credentials. This option is only used in the legacy retry mode.

• :retry_max_delay (Integer) — default: 0 —

  The maximum number of seconds to delay between retries (0 for no limit) used by the default backoff function. This option is only used in the legacy retry mode.

• :retry_mode (String) — default: "legacy" —

  Specifies which retry algorithm to use. Values are:

  • legacy - The pre-existing retry behavior. This is default value if no retry mode is provided.

  • standard - A standardized set of retry rules across the AWS SDKs. This includes support for retry quotas, which limit the number of unsuccessful retries a client can make.

  • adaptive - An experimental retry mode that includes all the functionality of standard mode along with automatic client side throttling. This is a provisional mode that may change behavior in the future.

• :sdk_ua_app_id (String) —

  A unique and opaque application ID that is appended to the User-Agent header as app/sdk_ua_app_id. It should have a maximum length of 50. This variable is sourced from environment variable AWS_SDK_UA_APP_ID or the shared config profile attribute sdk_ua_app_id.

• :secret_access_key (String)
• :session_token (String)
• :sigv4a_signing_region_set (Array) —

  A list of regions that should be signed with SigV4a signing. When not passed, a default :sigv4a_signing_region_set is searched for in the following locations:

  • Aws.config[:sigv4a_signing_region_set]
  • ENV['AWS_SIGV4A_SIGNING_REGION_SET']
  • ~/.aws/config
• :stub_responses (Boolean) — default: false —

  Causes the client to return stubbed responses. By default fake responses are generated and returned. You can specify the response data to return or errors to raise by calling ClientStubs#stub_responses. See ClientStubs for more information.

  Please note When response stubbing is enabled, no HTTP requests are made, and retries are disabled.

• :telemetry_provider (Aws::Telemetry::TelemetryProviderBase) — default: Aws::Telemetry::NoOpTelemetryProvider —

  Allows you to provide a telemetry provider, which is used to emit telemetry data. By default, uses NoOpTelemetryProvider which will not record or emit any telemetry data. The SDK supports the following telemetry providers:

  • OpenTelemetry (OTel) - To use the OTel provider, install and require the opentelemetry-sdk gem and then, pass in an instance of a Aws::Telemetry::OTelProvider for telemetry provider.
• :token_provider (Aws::TokenProvider) —

  A Bearer Token Provider. This can be an instance of any one of the following classes:

  • Aws::StaticTokenProvider - Used for configuring static, non-refreshing tokens.

  • Aws::SSOTokenProvider - Used for loading tokens from AWS SSO using an access token generated from aws login.

  When :token_provider is not configured directly, the Aws::TokenProviderChain will be used to search for tokens configured for your profile in shared configuration files.

• :use_dualstack_endpoint (Boolean) —

  When set to true, dualstack enabled endpoints (with .aws TLD) will be used if available.

• :use_fips_endpoint (Boolean) —

  When set to true, fips compatible endpoints will be used if available. When a fips region is used, the region is normalized and this config is set to true.

• :validate_params (Boolean) — default: true —

  When true, request parameters are validated before sending the request.

• :endpoint_provider (Aws::BedrockRuntime::EndpointProvider) —

  The endpoint provider used to resolve endpoints. Any object that responds to #resolve_endpoint(parameters) where parameters is a Struct similar to Aws::BedrockRuntime::EndpointParameters.

• :http_continue_timeout (Float) — default: 1 —

  The number of seconds to wait for a 100-continue response before sending the request body. This option has no effect unless the request has "Expect" header set to "100-continue". Defaults to nil which disables this behaviour. This value can safely be set per request on the session.

• :http_idle_timeout (Float) — default: 5 —

  The number of seconds a connection is allowed to sit idle before it is considered stale. Stale connections are closed and removed from the pool before making a request.

• :http_open_timeout (Float) — default: 15 —

  The default number of seconds to wait for response data. This value can safely be set per-request on the session.

• :http_proxy (URI::HTTP, String) —

  A proxy to send requests through. Formatted like 'http://proxy.com:123'.

• :http_read_timeout (Float) — default: 60 —

  The default number of seconds to wait for response data. This value can safely be set per-request on the session.

• :http_wire_trace (Boolean) — default: false —

  When true, HTTP debug output will be sent to the :logger.

• :on_chunk_received (Proc) —

  When a Proc object is provided, it will be used as callback when each chunk of the response body is received. It provides three arguments: the chunk, the number of bytes received, and the total number of bytes in the response (or nil if the server did not send a content-length).

• :on_chunk_sent (Proc) —

  When a Proc object is provided, it will be used as callback when each chunk of the request body is sent. It provides three arguments: the chunk, the number of bytes read from the body, and the total number of bytes in the body.

• :raise_response_errors (Boolean) — default: true —

  When true, response errors are raised.

• :ssl_ca_bundle (String) —

  Full path to the SSL certificate authority bundle file that should be used when verifying peer certificates. If you do not pass :ssl_ca_bundle or :ssl_ca_directory the the system default will be used if available.

• :ssl_ca_directory (String) —

  Full path of the directory that contains the unbundled SSL certificate authority files for verifying peer certificates. If you do not pass :ssl_ca_bundle or :ssl_ca_directory the the system default will be used if available.

• :ssl_ca_store (String) —

  Sets the X509::Store to verify peer certificate.

• :ssl_cert (OpenSSL::X509::Certificate) —

  Sets a client certificate when creating http connections.

• :ssl_key (OpenSSL::PKey) —

  Sets a client key when creating http connections.

• :ssl_timeout (Float) —

  Sets the SSL timeout in seconds

• :ssl_verify_peer (Boolean) — default: true —

  When true, SSL peer certificates are verified when establishing a connection.

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 455

def initialize(*args)
  super
end

Instance Method Details

#apply_guardrail(params = {}) ⇒ Types::ApplyGuardrailResponse

The action to apply a guardrail.

For troubleshooting some of the common errors you might encounter when using the ApplyGuardrail API, see Troubleshooting Amazon Bedrock API Error Codes in the Amazon Bedrock User Guide

Examples:

Request syntax with placeholder values

resp = client.apply_guardrail({
  guardrail_identifier: "GuardrailIdentifier", # required
  guardrail_version: "GuardrailVersion", # required
  source: "INPUT", # required, accepts INPUT, OUTPUT
  content: [ # required
    {
      text: {
        text: "String", # required
        qualifiers: ["grounding_source"], # accepts grounding_source, query, guard_content
      },
      image: {
        format: "png", # required, accepts png, jpeg
        source: { # required
          bytes: "data",
        },
      },
    },
  ],
})

Response structure

resp.usage.topic_policy_units #=> Integer
resp.usage.content_policy_units #=> Integer
resp.usage.word_policy_units #=> Integer
resp.usage.sensitive_information_policy_units #=> Integer
resp.usage.sensitive_information_policy_free_units #=> Integer
resp.usage.contextual_grounding_policy_units #=> Integer
resp.action #=> String, one of "NONE", "GUARDRAIL_INTERVENED"
resp.outputs #=> Array
resp.outputs[0].text #=> String
resp.assessments #=> Array
resp.assessments[0].topic_policy.topics #=> Array
resp.assessments[0].topic_policy.topics[0].name #=> String
resp.assessments[0].topic_policy.topics[0].type #=> String, one of "DENY"
resp.assessments[0].topic_policy.topics[0].action #=> String, one of "BLOCKED"
resp.assessments[0].content_policy.filters #=> Array
resp.assessments[0].content_policy.filters[0].type #=> String, one of "INSULTS", "HATE", "SEXUAL", "VIOLENCE", "MISCONDUCT", "PROMPT_ATTACK"
resp.assessments[0].content_policy.filters[0].confidence #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.assessments[0].content_policy.filters[0].filter_strength #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.assessments[0].content_policy.filters[0].action #=> String, one of "BLOCKED"
resp.assessments[0].word_policy.custom_words #=> Array
resp.assessments[0].word_policy.custom_words[0].match #=> String
resp.assessments[0].word_policy.custom_words[0].action #=> String, one of "BLOCKED"
resp.assessments[0].word_policy.managed_word_lists #=> Array
resp.assessments[0].word_policy.managed_word_lists[0].match #=> String
resp.assessments[0].word_policy.managed_word_lists[0].type #=> String, one of "PROFANITY"
resp.assessments[0].word_policy.managed_word_lists[0].action #=> String, one of "BLOCKED"
resp.assessments[0].sensitive_information_policy.pii_entities #=> Array
resp.assessments[0].sensitive_information_policy.pii_entities[0].match #=> String
resp.assessments[0].sensitive_information_policy.pii_entities[0].type #=> String, one of "ADDRESS", "AGE", "AWS_ACCESS_KEY", "AWS_SECRET_KEY", "CA_HEALTH_NUMBER", "CA_SOCIAL_INSURANCE_NUMBER", "CREDIT_DEBIT_CARD_CVV", "CREDIT_DEBIT_CARD_EXPIRY", "CREDIT_DEBIT_CARD_NUMBER", "DRIVER_ID", "EMAIL", "INTERNATIONAL_BANK_ACCOUNT_NUMBER", "IP_ADDRESS", "LICENSE_PLATE", "MAC_ADDRESS", "NAME", "PASSWORD", "PHONE", "PIN", "SWIFT_CODE", "UK_NATIONAL_HEALTH_SERVICE_NUMBER", "UK_NATIONAL_INSURANCE_NUMBER", "UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER", "URL", "USERNAME", "US_BANK_ACCOUNT_NUMBER", "US_BANK_ROUTING_NUMBER", "US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER", "US_PASSPORT_NUMBER", "US_SOCIAL_SECURITY_NUMBER", "VEHICLE_IDENTIFICATION_NUMBER"
resp.assessments[0].sensitive_information_policy.pii_entities[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.assessments[0].sensitive_information_policy.regexes #=> Array
resp.assessments[0].sensitive_information_policy.regexes[0].name #=> String
resp.assessments[0].sensitive_information_policy.regexes[0].match #=> String
resp.assessments[0].sensitive_information_policy.regexes[0].regex #=> String
resp.assessments[0].sensitive_information_policy.regexes[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.assessments[0].contextual_grounding_policy.filters #=> Array
resp.assessments[0].contextual_grounding_policy.filters[0].type #=> String, one of "GROUNDING", "RELEVANCE"
resp.assessments[0].contextual_grounding_policy.filters[0].threshold #=> Float
resp.assessments[0].contextual_grounding_policy.filters[0].score #=> Float
resp.assessments[0].contextual_grounding_policy.filters[0].action #=> String, one of "BLOCKED", "NONE"
resp.assessments[0].invocation_metrics.guardrail_processing_latency #=> Integer
resp.assessments[0].invocation_metrics.usage.topic_policy_units #=> Integer
resp.assessments[0].invocation_metrics.usage.content_policy_units #=> Integer
resp.assessments[0].invocation_metrics.usage.word_policy_units #=> Integer
resp.assessments[0].invocation_metrics.usage.sensitive_information_policy_units #=> Integer
resp.assessments[0].invocation_metrics.usage.sensitive_information_policy_free_units #=> Integer
resp.assessments[0].invocation_metrics.usage.contextual_grounding_policy_units #=> Integer
resp.assessments[0].invocation_metrics.guardrail_coverage.text_characters.guarded #=> Integer
resp.assessments[0].invocation_metrics.guardrail_coverage.text_characters.total #=> Integer
resp.assessments[0].invocation_metrics.guardrail_coverage.images.guarded #=> Integer
resp.assessments[0].invocation_metrics.guardrail_coverage.images.total #=> Integer
resp.guardrail_coverage.text_characters.guarded #=> Integer
resp.guardrail_coverage.text_characters.total #=> Integer
resp.guardrail_coverage.images.guarded #=> Integer
resp.guardrail_coverage.images.total #=> Integer

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :guardrail_identifier (required, String) —

  The guardrail identifier used in the request to apply the guardrail.

• :guardrail_version (required, String) —

  The guardrail version used in the request to apply the guardrail.

• :source (required, String) —

  The source of data used in the request to apply the guardrail.

• :content (required, Array<Types::GuardrailContentBlock>) —

  The content details used in the request to apply the guardrail.

Returns:

• (Types::ApplyGuardrailResponse) —

  Returns a response object which responds to the following methods:

  • #usage => Types::GuardrailUsage
  • #action => String
  • #outputs => Array<Types::GuardrailOutputContent>
  • #assessments => Array<Types::GuardrailAssessment>
  • #guardrail_coverage => Types::GuardrailCoverage

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 575

def apply_guardrail(params = {}, options = {})
  req = build_request(:apply_guardrail, params)
  req.send_request(options)
end

#converse(params = {}) ⇒ Types::ConverseResponse

Sends messages to the specified Amazon Bedrock model. Converse provides a consistent interface that works with all models that support messages. This allows you to write code once and use it with different models. If a model has unique inference parameters, you can also pass those unique parameters to the model.

Amazon Bedrock doesn't store any text, images, or documents that you provide as content. The data is only used to generate the response.

You can submit a prompt by including it in the messages field, specifying the modelId of a foundation model or inference profile to run inference on it, and including any other fields that are relevant to your use case.

You can also submit a prompt from Prompt management by specifying the ARN of the prompt version and including a map of variables to values in the promptVariables field. You can append more messages to the prompt by using the messages field. If you use a prompt from Prompt management, you can't include the following fields in the request: additionalModelRequestFields, inferenceConfig, system, or toolConfig. Instead, these fields must be defined through Prompt management. For more information, see Use a prompt from Prompt management.

For information about the Converse API, see Use the Converse API in the Amazon Bedrock User Guide. To use a guardrail, see Use a guardrail with the Converse API in the Amazon Bedrock User Guide. To use a tool with a model, see Tool use (Function calling) in the Amazon Bedrock User Guide

For example code, see Converse API examples in the Amazon Bedrock User Guide.

This operation requires permission for the bedrock:InvokeModel action.

To deny all inference access to resources that you specify in the modelId field, you need to deny access to the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream actions. Doing this also denies access to the resource through the base inference actions (InvokeModel and InvokeModelWithResponseStream). For more information see Deny access for inference on specific models.

For troubleshooting some of the common errors you might encounter when using the Converse API, see Troubleshooting Amazon Bedrock API Error Codes in the Amazon Bedrock User Guide

Examples:

Request syntax with placeholder values

resp = client.converse({
  model_id: "ConversationalModelId", # required
  messages: [
    {
      role: "user", # required, accepts user, assistant
      content: [ # required
        {
          text: "String",
          image: {
            format: "png", # required, accepts png, jpeg, gif, webp
            source: { # required
              bytes: "data",
            },
          },
          document: {
            format: "pdf", # required, accepts pdf, csv, doc, docx, xls, xlsx, html, txt, md
            name: "DocumentBlockNameString", # required
            source: { # required
              bytes: "data",
            },
          },
          video: {
            format: "mkv", # required, accepts mkv, mov, mp4, webm, flv, mpeg, mpg, wmv, three_gp
            source: { # required
              bytes: "data",
              s3_location: {
                uri: "S3Uri", # required
                bucket_owner: "AccountId",
              },
            },
          },
          tool_use: {
            tool_use_id: "ToolUseId", # required
            name: "ToolName", # required
            input: { # required
            },
          },
          tool_result: {
            tool_use_id: "ToolUseId", # required
            content: [ # required
              {
                json: {
                },
                text: "String",
                image: {
                  format: "png", # required, accepts png, jpeg, gif, webp
                  source: { # required
                    bytes: "data",
                  },
                },
                document: {
                  format: "pdf", # required, accepts pdf, csv, doc, docx, xls, xlsx, html, txt, md
                  name: "DocumentBlockNameString", # required
                  source: { # required
                    bytes: "data",
                  },
                },
                video: {
                  format: "mkv", # required, accepts mkv, mov, mp4, webm, flv, mpeg, mpg, wmv, three_gp
                  source: { # required
                    bytes: "data",
                    s3_location: {
                      uri: "S3Uri", # required
                      bucket_owner: "AccountId",
                    },
                  },
                },
              },
            ],
            status: "success", # accepts success, error
          },
          guard_content: {
            text: {
              text: "String", # required
              qualifiers: ["grounding_source"], # accepts grounding_source, query, guard_content
            },
            image: {
              format: "png", # required, accepts png, jpeg
              source: { # required
                bytes: "data",
              },
            },
          },
        },
      ],
    },
  ],
  system: [
    {
      text: "NonEmptyString",
      guard_content: {
        text: {
          text: "String", # required
          qualifiers: ["grounding_source"], # accepts grounding_source, query, guard_content
        },
        image: {
          format: "png", # required, accepts png, jpeg
          source: { # required
            bytes: "data",
          },
        },
      },
    },
  ],
  inference_config: {
    max_tokens: 1,
    temperature: 1.0,
    top_p: 1.0,
    stop_sequences: ["NonEmptyString"],
  },
  tool_config: {
    tools: [ # required
      {
        tool_spec: {
          name: "ToolName", # required
          description: "NonEmptyString",
          input_schema: { # required
            json: {
            },
          },
        },
      },
    ],
    tool_choice: {
      auto: {
      },
      any: {
      },
      tool: {
        name: "ToolName", # required
      },
    },
  },
  guardrail_config: {
    guardrail_identifier: "GuardrailIdentifier", # required
    guardrail_version: "GuardrailVersion", # required
    trace: "enabled", # accepts enabled, disabled
  },
  additional_model_request_fields: {
  },
  prompt_variables: {
    "String" => {
      text: "String",
    },
  },
  additional_model_response_field_paths: ["ConverseRequestAdditionalModelResponseFieldPathsListMemberString"],
  request_metadata: {
    "RequestMetadataKeyString" => "RequestMetadataValueString",
  },
  performance_config: {
    latency: "standard", # accepts standard, optimized
  },
})

Response structure

resp.output.message.role #=> String, one of "user", "assistant"
resp.output.message.content #=> Array
resp.output.message.content[0].text #=> String
resp.output.message.content[0].image.format #=> String, one of "png", "jpeg", "gif", "webp"
resp.output.message.content[0].image.source.bytes #=> String
resp.output.message.content[0].document.format #=> String, one of "pdf", "csv", "doc", "docx", "xls", "xlsx", "html", "txt", "md"
resp.output.message.content[0].document.name #=> String
resp.output.message.content[0].document.source.bytes #=> String
resp.output.message.content[0].video.format #=> String, one of "mkv", "mov", "mp4", "webm", "flv", "mpeg", "mpg", "wmv", "three_gp"
resp.output.message.content[0].video.source.bytes #=> String
resp.output.message.content[0].video.source.s3_location.uri #=> String
resp.output.message.content[0].video.source.s3_location.bucket_owner #=> String
resp.output.message.content[0].tool_use.tool_use_id #=> String
resp.output.message.content[0].tool_use.name #=> String
resp.output.message.content[0].tool_result.tool_use_id #=> String
resp.output.message.content[0].tool_result.content #=> Array
resp.output.message.content[0].tool_result.content[0].text #=> String
resp.output.message.content[0].tool_result.content[0].image.format #=> String, one of "png", "jpeg", "gif", "webp"
resp.output.message.content[0].tool_result.content[0].image.source.bytes #=> String
resp.output.message.content[0].tool_result.content[0].document.format #=> String, one of "pdf", "csv", "doc", "docx", "xls", "xlsx", "html", "txt", "md"
resp.output.message.content[0].tool_result.content[0].document.name #=> String
resp.output.message.content[0].tool_result.content[0].document.source.bytes #=> String
resp.output.message.content[0].tool_result.content[0].video.format #=> String, one of "mkv", "mov", "mp4", "webm", "flv", "mpeg", "mpg", "wmv", "three_gp"
resp.output.message.content[0].tool_result.content[0].video.source.bytes #=> String
resp.output.message.content[0].tool_result.content[0].video.source.s3_location.uri #=> String
resp.output.message.content[0].tool_result.content[0].video.source.s3_location.bucket_owner #=> String
resp.output.message.content[0].tool_result.status #=> String, one of "success", "error"
resp.output.message.content[0].guard_content.text.text #=> String
resp.output.message.content[0].guard_content.text.qualifiers #=> Array
resp.output.message.content[0].guard_content.text.qualifiers[0] #=> String, one of "grounding_source", "query", "guard_content"
resp.output.message.content[0].guard_content.image.format #=> String, one of "png", "jpeg"
resp.output.message.content[0].guard_content.image.source.bytes #=> String
resp.stop_reason #=> String, one of "end_turn", "tool_use", "max_tokens", "stop_sequence", "guardrail_intervened", "content_filtered"
resp.usage.input_tokens #=> Integer
resp.usage.output_tokens #=> Integer
resp.usage.total_tokens #=> Integer
resp.metrics.latency_ms #=> Integer
resp.trace.guardrail.model_output #=> Array
resp.trace.guardrail.model_output[0] #=> String
resp.trace.guardrail.input_assessment #=> Hash
resp.trace.guardrail.input_assessment["String"].topic_policy.topics #=> Array
resp.trace.guardrail.input_assessment["String"].topic_policy.topics[0].name #=> String
resp.trace.guardrail.input_assessment["String"].topic_policy.topics[0].type #=> String, one of "DENY"
resp.trace.guardrail.input_assessment["String"].topic_policy.topics[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.input_assessment["String"].content_policy.filters #=> Array
resp.trace.guardrail.input_assessment["String"].content_policy.filters[0].type #=> String, one of "INSULTS", "HATE", "SEXUAL", "VIOLENCE", "MISCONDUCT", "PROMPT_ATTACK"
resp.trace.guardrail.input_assessment["String"].content_policy.filters[0].confidence #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.trace.guardrail.input_assessment["String"].content_policy.filters[0].filter_strength #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.trace.guardrail.input_assessment["String"].content_policy.filters[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.input_assessment["String"].word_policy.custom_words #=> Array
resp.trace.guardrail.input_assessment["String"].word_policy.custom_words[0].match #=> String
resp.trace.guardrail.input_assessment["String"].word_policy.custom_words[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists #=> Array
resp.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].match #=> String
resp.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].type #=> String, one of "PROFANITY"
resp.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities #=> Array
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].match #=> String
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].type #=> String, one of "ADDRESS", "AGE", "AWS_ACCESS_KEY", "AWS_SECRET_KEY", "CA_HEALTH_NUMBER", "CA_SOCIAL_INSURANCE_NUMBER", "CREDIT_DEBIT_CARD_CVV", "CREDIT_DEBIT_CARD_EXPIRY", "CREDIT_DEBIT_CARD_NUMBER", "DRIVER_ID", "EMAIL", "INTERNATIONAL_BANK_ACCOUNT_NUMBER", "IP_ADDRESS", "LICENSE_PLATE", "MAC_ADDRESS", "NAME", "PASSWORD", "PHONE", "PIN", "SWIFT_CODE", "UK_NATIONAL_HEALTH_SERVICE_NUMBER", "UK_NATIONAL_INSURANCE_NUMBER", "UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER", "URL", "USERNAME", "US_BANK_ACCOUNT_NUMBER", "US_BANK_ROUTING_NUMBER", "US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER", "US_PASSPORT_NUMBER", "US_SOCIAL_SECURITY_NUMBER", "VEHICLE_IDENTIFICATION_NUMBER"
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes #=> Array
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].name #=> String
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].match #=> String
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].regex #=> String
resp.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters #=> Array
resp.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].type #=> String, one of "GROUNDING", "RELEVANCE"
resp.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].threshold #=> Float
resp.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].score #=> Float
resp.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].action #=> String, one of "BLOCKED", "NONE"
resp.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_processing_latency #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.topic_policy_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.content_policy_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.word_policy_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.sensitive_information_policy_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.sensitive_information_policy_free_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.usage.contextual_grounding_policy_units #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.text_characters.guarded #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.text_characters.total #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.images.guarded #=> Integer
resp.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.images.total #=> Integer
resp.trace.guardrail.output_assessments #=> Hash
resp.trace.guardrail.output_assessments["String"] #=> Array
resp.trace.guardrail.output_assessments["String"][0].topic_policy.topics #=> Array
resp.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].name #=> String
resp.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].type #=> String, one of "DENY"
resp.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].content_policy.filters #=> Array
resp.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].type #=> String, one of "INSULTS", "HATE", "SEXUAL", "VIOLENCE", "MISCONDUCT", "PROMPT_ATTACK"
resp.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].confidence #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].filter_strength #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
resp.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].word_policy.custom_words #=> Array
resp.trace.guardrail.output_assessments["String"][0].word_policy.custom_words[0].match #=> String
resp.trace.guardrail.output_assessments["String"][0].word_policy.custom_words[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists #=> Array
resp.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].match #=> String
resp.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].type #=> String, one of "PROFANITY"
resp.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].action #=> String, one of "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities #=> Array
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].match #=> String
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].type #=> String, one of "ADDRESS", "AGE", "AWS_ACCESS_KEY", "AWS_SECRET_KEY", "CA_HEALTH_NUMBER", "CA_SOCIAL_INSURANCE_NUMBER", "CREDIT_DEBIT_CARD_CVV", "CREDIT_DEBIT_CARD_EXPIRY", "CREDIT_DEBIT_CARD_NUMBER", "DRIVER_ID", "EMAIL", "INTERNATIONAL_BANK_ACCOUNT_NUMBER", "IP_ADDRESS", "LICENSE_PLATE", "MAC_ADDRESS", "NAME", "PASSWORD", "PHONE", "PIN", "SWIFT_CODE", "UK_NATIONAL_HEALTH_SERVICE_NUMBER", "UK_NATIONAL_INSURANCE_NUMBER", "UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER", "URL", "USERNAME", "US_BANK_ACCOUNT_NUMBER", "US_BANK_ROUTING_NUMBER", "US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER", "US_PASSPORT_NUMBER", "US_SOCIAL_SECURITY_NUMBER", "VEHICLE_IDENTIFICATION_NUMBER"
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes #=> Array
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].name #=> String
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].match #=> String
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].regex #=> String
resp.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
resp.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters #=> Array
resp.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].type #=> String, one of "GROUNDING", "RELEVANCE"
resp.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].threshold #=> Float
resp.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].score #=> Float
resp.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].action #=> String, one of "BLOCKED", "NONE"
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_processing_latency #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.topic_policy_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.content_policy_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.word_policy_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.sensitive_information_policy_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.sensitive_information_policy_free_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.contextual_grounding_policy_units #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.text_characters.guarded #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.text_characters.total #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.images.guarded #=> Integer
resp.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.images.total #=> Integer
resp.trace.prompt_router.invoked_model_id #=> String
resp.performance_config.latency #=> String, one of "standard", "optimized"

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :model_id (required, String) —

  Specifies the model or throughput with which to run inference, or the prompt resource to use in inference. The value depends on the resource that you use:

  • If you use a base model, specify the model ID or its ARN. For a list of model IDs for base models, see Amazon Bedrock base model IDs (on-demand throughput) in the Amazon Bedrock User Guide.

  • If you use an inference profile, specify the inference profile ID or its ARN. For a list of inference profile IDs, see Supported Regions and models for cross-region inference in the Amazon Bedrock User Guide.

  • If you use a provisioned model, specify the ARN of the Provisioned Throughput. For more information, see Run inference using a Provisioned Throughput in the Amazon Bedrock User Guide.

  • If you use a custom model, first purchase Provisioned Throughput for it. Then specify the ARN of the resulting provisioned model. For more information, see Use a custom model in Amazon Bedrock in the Amazon Bedrock User Guide.

  • To include a prompt that was defined in Prompt management, specify the ARN of the prompt version to use.

  The Converse API doesn't support imported models.

• :messages (Array<Types::Message>) —

  The messages that you want to send to the model.

• :system (Array<Types::SystemContentBlock>) —

  A prompt that provides instructions or context to the model about the task it should perform, or the persona it should adopt during the conversation.

• :inference_config (Types::InferenceConfiguration) —

  Inference parameters to pass to the model. Converse and ConverseStream support a base set of inference parameters. If you need to pass additional parameters that the model supports, use the additionalModelRequestFields request field.

• :tool_config (Types::ToolConfiguration) —

  Configuration information for the tools that the model can use when generating a response.

  For information about models that support tool use, see Supported models and model features.

• :guardrail_config (Types::GuardrailConfiguration) —

  Configuration information for a guardrail that you want to use in the request. If you include guardContent blocks in the content field in the messages field, the guardrail operates only on those messages. If you include no guardContent blocks, the guardrail operates on all messages in the request body and in any included prompt resource.

• :additional_model_request_fields (Hash, Array, String, Numeric, Boolean) —

  Additional inference parameters that the model supports, beyond the base set of inference parameters that Converse and ConverseStream support in the inferenceConfig field. For more information, see Model parameters.

  Document type used to carry open content (Hash,Array,String,Numeric,Boolean). A document type value is serialized using the same format as its surroundings and requires no additional encoding or escaping.

• :prompt_variables (Hash<String,Types::PromptVariableValues>) —

  Contains a map of variables in a prompt from Prompt management to objects containing the values to fill in for them when running model invocation. This field is ignored if you don't specify a prompt resource in the modelId field.

• :additional_model_response_field_paths (Array<String>) —

  Additional model parameters field paths to return in the response. Converse and ConverseStream return the requested fields as a JSON Pointer object in the additionalModelResponseFields field. The following is example JSON for additionalModelResponseFieldPaths.

  [ "/stop_sequence" ]

  For information about the JSON Pointer syntax, see the Internet Engineering Task Force (IETF) documentation.

  Converse and ConverseStream reject an empty JSON Pointer or incorrectly structured JSON Pointer with a 400 error code. if the JSON Pointer is valid, but the requested field is not in the model response, it is ignored by Converse.

• :request_metadata (Hash<String,String>) —

  Key-value pairs that you can use to filter invocation logs.

• :performance_config (Types::PerformanceConfiguration) —

  Model performance settings for the request.

Returns:

• (Types::ConverseResponse) —

  Returns a response object which responds to the following methods:

  • #output => Types::ConverseOutput
  • #stop_reason => String
  • #usage => Types::TokenUsage
  • #metrics => Types::ConverseMetrics
  • #additional_model_response_fields => Hash,Array,String,Numeric,Boolean
  • #trace => Types::ConverseTrace
  • #performance_config => Types::PerformanceConfiguration

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 1051

def converse(params = {}, options = {})
  req = build_request(:converse, params)
  req.send_request(options)
end

#converse_stream(params = {}) ⇒ Types::ConverseStreamResponse

Sends messages to the specified Amazon Bedrock model and returns the response in a stream. ConverseStream provides a consistent API that works with all Amazon Bedrock models that support messages. This allows you to write code once and use it with different models. Should a model have unique inference parameters, you can also pass those unique parameters to the model.

To find out if a model supports streaming, call GetFoundationModel and check the responseStreamingSupported field in the response.

The CLI doesn't support streaming operations in Amazon Bedrock, including ConverseStream.

Amazon Bedrock doesn't store any text, images, or documents that you provide as content. The data is only used to generate the response.

You can submit a prompt by including it in the messages field, specifying the modelId of a foundation model or inference profile to run inference on it, and including any other fields that are relevant to your use case.

You can also submit a prompt from Prompt management by specifying the ARN of the prompt version and including a map of variables to values in the promptVariables field. You can append more messages to the prompt by using the messages field. If you use a prompt from Prompt management, you can't include the following fields in the request: additionalModelRequestFields, inferenceConfig, system, or toolConfig. Instead, these fields must be defined through Prompt management. For more information, see Use a prompt from Prompt management.

For information about the Converse API, see Use the Converse API in the Amazon Bedrock User Guide. To use a guardrail, see Use a guardrail with the Converse API in the Amazon Bedrock User Guide. To use a tool with a model, see Tool use (Function calling) in the Amazon Bedrock User Guide

For example code, see Conversation streaming example in the Amazon Bedrock User Guide.

This operation requires permission for the bedrock:InvokeModelWithResponseStream action.

To deny all inference access to resources that you specify in the modelId field, you need to deny access to the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream actions. Doing this also denies access to the resource through the base inference actions (InvokeModel and InvokeModelWithResponseStream). For more information see Deny access for inference on specific models.

For troubleshooting some of the common errors you might encounter when using the ConverseStream API, see Troubleshooting Amazon Bedrock API Error Codes in the Amazon Bedrock User Guide

Examples:

EventStream Operation Example

You can process the event once it arrives immediately, or wait until the
full response is complete and iterate through the eventstream enumerator.

To interact with event immediately, you need to register #converse_stream
with callbacks. Callbacks can be registered for specific events or for all
events, including error events.

Callbacks can be passed into the `:event_stream_handler` option or within a
block statement attached to the #converse_stream call directly. Hybrid
pattern of both is also supported.

`:event_stream_handler` option takes in either a Proc object or
Aws::BedrockRuntime::EventStreams::ConverseStreamOutput object.

Usage pattern a): Callbacks with a block attached to #converse_stream
  Example for registering callbacks for all event types and an error event

  client.converse_stream( # params input# ) do |stream|
    stream.on_error_event do |event|
      # catch unmodeled error event in the stream
      raise event
      # => Aws::Errors::EventError
      # event.event_type => :error
      # event.error_code => String
      # event.error_message => String
    end

    stream.on_event do |event|
      # process all events arrive
      puts event.event_type
      ...
    end

  end

Usage pattern b): Pass in `:event_stream_handler` for #converse_stream

  1) Create a Aws::BedrockRuntime::EventStreams::ConverseStreamOutput object
  Example for registering callbacks with specific events

    handler = Aws::BedrockRuntime::EventStreams::ConverseStreamOutput.new
    handler.on_message_start_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStart
    end
    handler.on_content_block_start_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStart
    end
    handler.on_content_block_delta_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockDelta
    end
    handler.on_content_block_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStop
    end
    handler.on_message_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStop
    end
    handler.on_metadata_event do |event|
      event # => Aws::BedrockRuntime::Types::metadata
    end
    handler.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    handler.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    handler.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    handler.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    handler.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end

  client.converse_stream( # params input #, event_stream_handler: handler)

  2) Use a Ruby Proc object
  Example for registering callbacks with specific events

  handler = Proc.new do |stream|
    stream.on_message_start_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStart
    end
    stream.on_content_block_start_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStart
    end
    stream.on_content_block_delta_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockDelta
    end
    stream.on_content_block_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStop
    end
    stream.on_message_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStop
    end
    stream.on_metadata_event do |event|
      event # => Aws::BedrockRuntime::Types::metadata
    end
    stream.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    stream.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    stream.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    stream.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    stream.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end
  end

  client.converse_stream( # params input #, event_stream_handler: handler)

Usage pattern c): Hybrid pattern of a) and b)

    handler = Aws::BedrockRuntime::EventStreams::ConverseStreamOutput.new
    handler.on_message_start_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStart
    end
    handler.on_content_block_start_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStart
    end
    handler.on_content_block_delta_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockDelta
    end
    handler.on_content_block_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::contentBlockStop
    end
    handler.on_message_stop_event do |event|
      event # => Aws::BedrockRuntime::Types::messageStop
    end
    handler.on_metadata_event do |event|
      event # => Aws::BedrockRuntime::Types::metadata
    end
    handler.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    handler.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    handler.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    handler.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    handler.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end

  client.converse_stream( # params input #, event_stream_handler: handler) do |stream|
    stream.on_error_event do |event|
      # catch unmodeled error event in the stream
      raise event
      # => Aws::Errors::EventError
      # event.event_type => :error
      # event.error_code => String
      # event.error_message => String
    end
  end

You can also iterate through events after the response complete.

Events are available at resp.stream # => Enumerator
For parameter input example, please refer to following request syntax

Request syntax with placeholder values

resp = client.converse_stream({
  model_id: "ConversationalModelId", # required
  messages: [
    {
      role: "user", # required, accepts user, assistant
      content: [ # required
        {
          text: "String",
          image: {
            format: "png", # required, accepts png, jpeg, gif, webp
            source: { # required
              bytes: "data",
            },
          },
          document: {
            format: "pdf", # required, accepts pdf, csv, doc, docx, xls, xlsx, html, txt, md
            name: "DocumentBlockNameString", # required
            source: { # required
              bytes: "data",
            },
          },
          video: {
            format: "mkv", # required, accepts mkv, mov, mp4, webm, flv, mpeg, mpg, wmv, three_gp
            source: { # required
              bytes: "data",
              s3_location: {
                uri: "S3Uri", # required
                bucket_owner: "AccountId",
              },
            },
          },
          tool_use: {
            tool_use_id: "ToolUseId", # required
            name: "ToolName", # required
            input: { # required
            },
          },
          tool_result: {
            tool_use_id: "ToolUseId", # required
            content: [ # required
              {
                json: {
                },
                text: "String",
                image: {
                  format: "png", # required, accepts png, jpeg, gif, webp
                  source: { # required
                    bytes: "data",
                  },
                },
                document: {
                  format: "pdf", # required, accepts pdf, csv, doc, docx, xls, xlsx, html, txt, md
                  name: "DocumentBlockNameString", # required
                  source: { # required
                    bytes: "data",
                  },
                },
                video: {
                  format: "mkv", # required, accepts mkv, mov, mp4, webm, flv, mpeg, mpg, wmv, three_gp
                  source: { # required
                    bytes: "data",
                    s3_location: {
                      uri: "S3Uri", # required
                      bucket_owner: "AccountId",
                    },
                  },
                },
              },
            ],
            status: "success", # accepts success, error
          },
          guard_content: {
            text: {
              text: "String", # required
              qualifiers: ["grounding_source"], # accepts grounding_source, query, guard_content
            },
            image: {
              format: "png", # required, accepts png, jpeg
              source: { # required
                bytes: "data",
              },
            },
          },
        },
      ],
    },
  ],
  system: [
    {
      text: "NonEmptyString",
      guard_content: {
        text: {
          text: "String", # required
          qualifiers: ["grounding_source"], # accepts grounding_source, query, guard_content
        },
        image: {
          format: "png", # required, accepts png, jpeg
          source: { # required
            bytes: "data",
          },
        },
      },
    },
  ],
  inference_config: {
    max_tokens: 1,
    temperature: 1.0,
    top_p: 1.0,
    stop_sequences: ["NonEmptyString"],
  },
  tool_config: {
    tools: [ # required
      {
        tool_spec: {
          name: "ToolName", # required
          description: "NonEmptyString",
          input_schema: { # required
            json: {
            },
          },
        },
      },
    ],
    tool_choice: {
      auto: {
      },
      any: {
      },
      tool: {
        name: "ToolName", # required
      },
    },
  },
  guardrail_config: {
    guardrail_identifier: "GuardrailIdentifier", # required
    guardrail_version: "GuardrailVersion", # required
    trace: "enabled", # accepts enabled, disabled
    stream_processing_mode: "sync", # accepts sync, async
  },
  additional_model_request_fields: {
  },
  prompt_variables: {
    "String" => {
      text: "String",
    },
  },
  additional_model_response_field_paths: ["ConverseStreamRequestAdditionalModelResponseFieldPathsListMemberString"],
  request_metadata: {
    "RequestMetadataKeyString" => "RequestMetadataValueString",
  },
  performance_config: {
    latency: "standard", # accepts standard, optimized
  },
})

Response structure

All events are available at resp.stream:
resp.stream #=> Enumerator
resp.stream.event_types #=> [:message_start, :content_block_start, :content_block_delta, :content_block_stop, :message_stop, :metadata, :internal_server_exception, :model_stream_error_exception, :validation_exception, :throttling_exception, :service_unavailable_exception]

For :message_start event available at #on_message_start_event callback and response eventstream enumerator:
event.role #=> String, one of "user", "assistant"

For :content_block_start event available at #on_content_block_start_event callback and response eventstream enumerator:
event.start.tool_use.tool_use_id #=> String
event.start.tool_use.name #=> String
event.content_block_index #=> Integer

For :content_block_delta event available at #on_content_block_delta_event callback and response eventstream enumerator:
event.delta.text #=> String
event.delta.tool_use.input #=> String
event.content_block_index #=> Integer

For :content_block_stop event available at #on_content_block_stop_event callback and response eventstream enumerator:
event.content_block_index #=> Integer

For :message_stop event available at #on_message_stop_event callback and response eventstream enumerator:
event.stop_reason #=> String, one of "end_turn", "tool_use", "max_tokens", "stop_sequence", "guardrail_intervened", "content_filtered"

For :metadata event available at #on_metadata_event callback and response eventstream enumerator:
event.usage.input_tokens #=> Integer
event.usage.output_tokens #=> Integer
event.usage.total_tokens #=> Integer
event.metrics.latency_ms #=> Integer
event.trace.guardrail.model_output #=> Array
event.trace.guardrail.model_output[0] #=> String
event.trace.guardrail.input_assessment #=> Hash
event.trace.guardrail.input_assessment["String"].topic_policy.topics #=> Array
event.trace.guardrail.input_assessment["String"].topic_policy.topics[0].name #=> String
event.trace.guardrail.input_assessment["String"].topic_policy.topics[0].type #=> String, one of "DENY"
event.trace.guardrail.input_assessment["String"].topic_policy.topics[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.input_assessment["String"].content_policy.filters #=> Array
event.trace.guardrail.input_assessment["String"].content_policy.filters[0].type #=> String, one of "INSULTS", "HATE", "SEXUAL", "VIOLENCE", "MISCONDUCT", "PROMPT_ATTACK"
event.trace.guardrail.input_assessment["String"].content_policy.filters[0].confidence #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
event.trace.guardrail.input_assessment["String"].content_policy.filters[0].filter_strength #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
event.trace.guardrail.input_assessment["String"].content_policy.filters[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.input_assessment["String"].word_policy.custom_words #=> Array
event.trace.guardrail.input_assessment["String"].word_policy.custom_words[0].match #=> String
event.trace.guardrail.input_assessment["String"].word_policy.custom_words[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists #=> Array
event.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].match #=> String
event.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].type #=> String, one of "PROFANITY"
event.trace.guardrail.input_assessment["String"].word_policy.managed_word_lists[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities #=> Array
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].match #=> String
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].type #=> String, one of "ADDRESS", "AGE", "AWS_ACCESS_KEY", "AWS_SECRET_KEY", "CA_HEALTH_NUMBER", "CA_SOCIAL_INSURANCE_NUMBER", "CREDIT_DEBIT_CARD_CVV", "CREDIT_DEBIT_CARD_EXPIRY", "CREDIT_DEBIT_CARD_NUMBER", "DRIVER_ID", "EMAIL", "INTERNATIONAL_BANK_ACCOUNT_NUMBER", "IP_ADDRESS", "LICENSE_PLATE", "MAC_ADDRESS", "NAME", "PASSWORD", "PHONE", "PIN", "SWIFT_CODE", "UK_NATIONAL_HEALTH_SERVICE_NUMBER", "UK_NATIONAL_INSURANCE_NUMBER", "UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER", "URL", "USERNAME", "US_BANK_ACCOUNT_NUMBER", "US_BANK_ROUTING_NUMBER", "US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER", "US_PASSPORT_NUMBER", "US_SOCIAL_SECURITY_NUMBER", "VEHICLE_IDENTIFICATION_NUMBER"
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.pii_entities[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes #=> Array
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].name #=> String
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].match #=> String
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].regex #=> String
event.trace.guardrail.input_assessment["String"].sensitive_information_policy.regexes[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
event.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters #=> Array
event.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].type #=> String, one of "GROUNDING", "RELEVANCE"
event.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].threshold #=> Float
event.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].score #=> Float
event.trace.guardrail.input_assessment["String"].contextual_grounding_policy.filters[0].action #=> String, one of "BLOCKED", "NONE"
event.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_processing_latency #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.topic_policy_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.content_policy_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.word_policy_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.sensitive_information_policy_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.sensitive_information_policy_free_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.usage.contextual_grounding_policy_units #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.text_characters.guarded #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.text_characters.total #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.images.guarded #=> Integer
event.trace.guardrail.input_assessment["String"].invocation_metrics.guardrail_coverage.images.total #=> Integer
event.trace.guardrail.output_assessments #=> Hash
event.trace.guardrail.output_assessments["String"] #=> Array
event.trace.guardrail.output_assessments["String"][0].topic_policy.topics #=> Array
event.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].name #=> String
event.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].type #=> String, one of "DENY"
event.trace.guardrail.output_assessments["String"][0].topic_policy.topics[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].content_policy.filters #=> Array
event.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].type #=> String, one of "INSULTS", "HATE", "SEXUAL", "VIOLENCE", "MISCONDUCT", "PROMPT_ATTACK"
event.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].confidence #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
event.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].filter_strength #=> String, one of "NONE", "LOW", "MEDIUM", "HIGH"
event.trace.guardrail.output_assessments["String"][0].content_policy.filters[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].word_policy.custom_words #=> Array
event.trace.guardrail.output_assessments["String"][0].word_policy.custom_words[0].match #=> String
event.trace.guardrail.output_assessments["String"][0].word_policy.custom_words[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists #=> Array
event.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].match #=> String
event.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].type #=> String, one of "PROFANITY"
event.trace.guardrail.output_assessments["String"][0].word_policy.managed_word_lists[0].action #=> String, one of "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities #=> Array
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].match #=> String
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].type #=> String, one of "ADDRESS", "AGE", "AWS_ACCESS_KEY", "AWS_SECRET_KEY", "CA_HEALTH_NUMBER", "CA_SOCIAL_INSURANCE_NUMBER", "CREDIT_DEBIT_CARD_CVV", "CREDIT_DEBIT_CARD_EXPIRY", "CREDIT_DEBIT_CARD_NUMBER", "DRIVER_ID", "EMAIL", "INTERNATIONAL_BANK_ACCOUNT_NUMBER", "IP_ADDRESS", "LICENSE_PLATE", "MAC_ADDRESS", "NAME", "PASSWORD", "PHONE", "PIN", "SWIFT_CODE", "UK_NATIONAL_HEALTH_SERVICE_NUMBER", "UK_NATIONAL_INSURANCE_NUMBER", "UK_UNIQUE_TAXPAYER_REFERENCE_NUMBER", "URL", "USERNAME", "US_BANK_ACCOUNT_NUMBER", "US_BANK_ROUTING_NUMBER", "US_INDIVIDUAL_TAX_IDENTIFICATION_NUMBER", "US_PASSPORT_NUMBER", "US_SOCIAL_SECURITY_NUMBER", "VEHICLE_IDENTIFICATION_NUMBER"
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.pii_entities[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes #=> Array
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].name #=> String
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].match #=> String
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].regex #=> String
event.trace.guardrail.output_assessments["String"][0].sensitive_information_policy.regexes[0].action #=> String, one of "ANONYMIZED", "BLOCKED"
event.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters #=> Array
event.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].type #=> String, one of "GROUNDING", "RELEVANCE"
event.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].threshold #=> Float
event.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].score #=> Float
event.trace.guardrail.output_assessments["String"][0].contextual_grounding_policy.filters[0].action #=> String, one of "BLOCKED", "NONE"
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_processing_latency #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.topic_policy_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.content_policy_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.word_policy_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.sensitive_information_policy_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.sensitive_information_policy_free_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.usage.contextual_grounding_policy_units #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.text_characters.guarded #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.text_characters.total #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.images.guarded #=> Integer
event.trace.guardrail.output_assessments["String"][0].invocation_metrics.guardrail_coverage.images.total #=> Integer
event.trace.prompt_router.invoked_model_id #=> String
event.performance_config.latency #=> String, one of "standard", "optimized"

For :internal_server_exception event available at #on_internal_server_exception_event callback and response eventstream enumerator:
event.message #=> String

For :model_stream_error_exception event available at #on_model_stream_error_exception_event callback and response eventstream enumerator:
event.message #=> String
event.original_status_code #=> Integer
event.original_message #=> String

For :validation_exception event available at #on_validation_exception_event callback and response eventstream enumerator:
event.message #=> String

For :throttling_exception event available at #on_throttling_exception_event callback and response eventstream enumerator:
event.message #=> String

For :service_unavailable_exception event available at #on_service_unavailable_exception_event callback and response eventstream enumerator:
event.message #=> String

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :model_id (required, String) —

  Specifies the model or throughput with which to run inference, or the prompt resource to use in inference. The value depends on the resource that you use:

  • If you use a base model, specify the model ID or its ARN. For a list of model IDs for base models, see Amazon Bedrock base model IDs (on-demand throughput) in the Amazon Bedrock User Guide.

  • If you use an inference profile, specify the inference profile ID or its ARN. For a list of inference profile IDs, see Supported Regions and models for cross-region inference in the Amazon Bedrock User Guide.

  • If you use a provisioned model, specify the ARN of the Provisioned Throughput. For more information, see Run inference using a Provisioned Throughput in the Amazon Bedrock User Guide.

  • If you use a custom model, first purchase Provisioned Throughput for it. Then specify the ARN of the resulting provisioned model. For more information, see Use a custom model in Amazon Bedrock in the Amazon Bedrock User Guide.

  • To include a prompt that was defined in Prompt management, specify the ARN of the prompt version to use.

  The Converse API doesn't support imported models.

• :messages (Array<Types::Message>) —

  The messages that you want to send to the model.

• :system (Array<Types::SystemContentBlock>) —

  A prompt that provides instructions or context to the model about the task it should perform, or the persona it should adopt during the conversation.

• :inference_config (Types::InferenceConfiguration) —

  Inference parameters to pass to the model. Converse and ConverseStream support a base set of inference parameters. If you need to pass additional parameters that the model supports, use the additionalModelRequestFields request field.

• :tool_config (Types::ToolConfiguration) —

  Configuration information for the tools that the model can use when generating a response.

  For information about models that support streaming tool use, see Supported models and model features.

• :guardrail_config (Types::GuardrailStreamConfiguration) —

  Configuration information for a guardrail that you want to use in the request. If you include guardContent blocks in the content field in the messages field, the guardrail operates only on those messages. If you include no guardContent blocks, the guardrail operates on all messages in the request body and in any included prompt resource.

• :additional_model_request_fields (Hash, Array, String, Numeric, Boolean) —

  Additional inference parameters that the model supports, beyond the base set of inference parameters that Converse and ConverseStream support in the inferenceConfig field. For more information, see Model parameters.

  Document type used to carry open content (Hash,Array,String,Numeric,Boolean). A document type value is serialized using the same format as its surroundings and requires no additional encoding or escaping.

• :prompt_variables (Hash<String,Types::PromptVariableValues>) —

  Contains a map of variables in a prompt from Prompt management to objects containing the values to fill in for them when running model invocation. This field is ignored if you don't specify a prompt resource in the modelId field.

• :additional_model_response_field_paths (Array<String>) —

  Additional model parameters field paths to return in the response. Converse and ConverseStream return the requested fields as a JSON Pointer object in the additionalModelResponseFields field. The following is example JSON for additionalModelResponseFieldPaths.

  [ "/stop_sequence" ]

  For information about the JSON Pointer syntax, see the Internet Engineering Task Force (IETF) documentation.

  Converse and ConverseStream reject an empty JSON Pointer or incorrectly structured JSON Pointer with a 400 error code. if the JSON Pointer is valid, but the requested field is not in the model response, it is ignored by Converse.

• :request_metadata (Hash<String,String>) —

  Key-value pairs that you can use to filter invocation logs.

• :performance_config (Types::PerformanceConfiguration) —

  Model performance settings for the request.

Yields:

• (event_stream_handler)

Returns:

• (Types::ConverseStreamResponse) —

  Returns a response object which responds to the following methods:

  • #stream => Types::ConverseStreamOutput

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 1714

def converse_stream(params = {}, options = {}, &block)
  params = params.dup
  event_stream_handler = case handler = params.delete(:event_stream_handler)
    when EventStreams::ConverseStreamOutput then handler
    when Proc then EventStreams::ConverseStreamOutput.new.tap(&handler)
    when nil then EventStreams::ConverseStreamOutput.new
    else
      msg = "expected :event_stream_handler to be a block or "\
            "instance of Aws::BedrockRuntime::EventStreams::ConverseStreamOutput"\
            ", got `#{handler.inspect}` instead"
      raise ArgumentError, msg
    end

  yield(event_stream_handler) if block_given?

  req = build_request(:converse_stream, params)

  req.context[:event_stream_handler] = event_stream_handler
  req.handlers.add(Aws::Binary::DecodeHandler, priority: 95)

  req.send_request(options, &block)
end

#get_async_invoke(params = {}) ⇒ Types::GetAsyncInvokeResponse

Retrieve information about an asynchronous invocation.

Examples:

Request syntax with placeholder values

resp = client.get_async_invoke({
  invocation_arn: "InvocationArn", # required
})

Response structure

resp.invocation_arn #=> String
resp.model_arn #=> String
resp.client_request_token #=> String
resp.status #=> String, one of "InProgress", "Completed", "Failed"
resp.failure_message #=> String
resp.submit_time #=> Time
resp.last_modified_time #=> Time
resp.end_time #=> Time
resp.output_data_config.s3_output_data_config.s3_uri #=> String
resp.output_data_config.s3_output_data_config.kms_key_id #=> String
resp.output_data_config.s3_output_data_config.bucket_owner #=> String

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :invocation_arn (required, String) —

  The invocation's ARN.

Returns:

• (Types::GetAsyncInvokeResponse) —

  Returns a response object which responds to the following methods:

  • #invocation_arn => String
  • #model_arn => String
  • #client_request_token => String
  • #status => String
  • #failure_message => String
  • #submit_time => Time
  • #last_modified_time => Time
  • #end_time => Time
  • #output_data_config => Types::AsyncInvokeOutputDataConfig

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 1778

def get_async_invoke(params = {}, options = {})
  req = build_request(:get_async_invoke, params)
  req.send_request(options)
end

#invoke_model(params = {}) ⇒ Types::InvokeModelResponse

Invokes the specified Amazon Bedrock model to run inference using the prompt and inference parameters provided in the request body. You use model inference to generate text, images, and embeddings.

For example code, see Invoke model code examples in the Amazon Bedrock User Guide.

This operation requires permission for the bedrock:InvokeModel action.

To deny all inference access to resources that you specify in the modelId field, you need to deny access to the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream actions. Doing this also denies access to the resource through the Converse API actions (Converse and ConverseStream). For more information see Deny access for inference on specific models.

For troubleshooting some of the common errors you might encounter when using the InvokeModel API, see Troubleshooting Amazon Bedrock API Error Codes in the Amazon Bedrock User Guide

Examples:

Request syntax with placeholder values

resp = client.invoke_model({
  body: "data",
  content_type: "MimeType",
  accept: "MimeType",
  model_id: "InvokeModelIdentifier", # required
  trace: "ENABLED", # accepts ENABLED, DISABLED
  guardrail_identifier: "GuardrailIdentifier",
  guardrail_version: "GuardrailVersion",
  performance_config_latency: "standard", # accepts standard, optimized
})

Response structure

resp.body #=> String
resp.content_type #=> String
resp.performance_config_latency #=> String, one of "standard", "optimized"

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :body (String, StringIO, File) —

  The prompt and inference parameters in the format specified in the contentType in the header. You must provide the body in JSON format. To see the format and content of the request and response bodies for different models, refer to Inference parameters. For more information, see Run inference in the Bedrock User Guide.

• :content_type (String) —

  The MIME type of the input data in the request. You must specify application/json.

• :accept (String) —

  The desired MIME type of the inference body in the response. The default value is application/json.

• :model_id (required, String) —

  The unique identifier of the model to invoke to run inference.

  The modelId to provide depends on the type of model or throughput that you use:

  • If you use a base model, specify the model ID or its ARN. For a list of model IDs for base models, see Amazon Bedrock base model IDs (on-demand throughput) in the Amazon Bedrock User Guide.

  • If you use an inference profile, specify the inference profile ID or its ARN. For a list of inference profile IDs, see Supported Regions and models for cross-region inference in the Amazon Bedrock User Guide.

  • If you use a provisioned model, specify the ARN of the Provisioned Throughput. For more information, see Run inference using a Provisioned Throughput in the Amazon Bedrock User Guide.

  • If you use a custom model, first purchase Provisioned Throughput for it. Then specify the ARN of the resulting provisioned model. For more information, see Use a custom model in Amazon Bedrock in the Amazon Bedrock User Guide.

  • If you use an imported model, specify the ARN of the imported model. You can get the model ARN from a successful call to CreateModelImportJob or from the Imported models page in the Amazon Bedrock console.

• :trace (String) —

  Specifies whether to enable or disable the Bedrock trace. If enabled, you can see the full Bedrock trace.

• :guardrail_identifier (String) —

  The unique identifier of the guardrail that you want to use. If you don't provide a value, no guardrail is applied to the invocation.

  An error will be thrown in the following situations.

  • You don't provide a guardrail identifier but you specify the amazon-bedrock-guardrailConfig field in the request body.

  • You enable the guardrail but the contentType isn't application/json.

  • You provide a guardrail identifier, but guardrailVersion isn't specified.

• :guardrail_version (String) —

  The version number for the guardrail. The value can also be DRAFT.

• :performance_config_latency (String) —

  Model performance settings for the request.

Returns:

• (Types::InvokeModelResponse) —

  Returns a response object which responds to the following methods:

  • #body => String
  • #content_type => String
  • #performance_config_latency => String

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 1923

def invoke_model(params = {}, options = {})
  req = build_request(:invoke_model, params)
  req.send_request(options)
end

#invoke_model_with_response_stream(params = {}) ⇒ Types::InvokeModelWithResponseStreamResponse

Invoke the specified Amazon Bedrock model to run inference using the prompt and inference parameters provided in the request body. The response is returned in a stream.

To see if a model supports streaming, call GetFoundationModel and check the responseStreamingSupported field in the response.

The CLI doesn't support streaming operations in Amazon Bedrock, including InvokeModelWithResponseStream.

For example code, see Invoke model with streaming code example in the Amazon Bedrock User Guide.

This operation requires permissions to perform the bedrock:InvokeModelWithResponseStream action.

To deny all inference access to resources that you specify in the modelId field, you need to deny access to the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream actions. Doing this also denies access to the resource through the Converse API actions (Converse and ConverseStream). For more information see Deny access for inference on specific models.

For troubleshooting some of the common errors you might encounter when using the InvokeModelWithResponseStream API, see Troubleshooting Amazon Bedrock API Error Codes in the Amazon Bedrock User Guide

Examples:

EventStream Operation Example

You can process the event once it arrives immediately, or wait until the
full response is complete and iterate through the eventstream enumerator.

To interact with event immediately, you need to register #invoke_model_with_response_stream
with callbacks. Callbacks can be registered for specific events or for all
events, including error events.

Callbacks can be passed into the `:event_stream_handler` option or within a
block statement attached to the #invoke_model_with_response_stream call directly. Hybrid
pattern of both is also supported.

`:event_stream_handler` option takes in either a Proc object or
Aws::BedrockRuntime::EventStreams::ResponseStream object.

Usage pattern a): Callbacks with a block attached to #invoke_model_with_response_stream
  Example for registering callbacks for all event types and an error event

  client.invoke_model_with_response_stream( # params input# ) do |stream|
    stream.on_error_event do |event|
      # catch unmodeled error event in the stream
      raise event
      # => Aws::Errors::EventError
      # event.event_type => :error
      # event.error_code => String
      # event.error_message => String
    end

    stream.on_event do |event|
      # process all events arrive
      puts event.event_type
      ...
    end

  end

Usage pattern b): Pass in `:event_stream_handler` for #invoke_model_with_response_stream

  1) Create a Aws::BedrockRuntime::EventStreams::ResponseStream object
  Example for registering callbacks with specific events

    handler = Aws::BedrockRuntime::EventStreams::ResponseStream.new
    handler.on_chunk_event do |event|
      event # => Aws::BedrockRuntime::Types::chunk
    end
    handler.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    handler.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    handler.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    handler.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    handler.on_model_timeout_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelTimeoutException
    end
    handler.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end

  client.invoke_model_with_response_stream( # params input #, event_stream_handler: handler)

  2) Use a Ruby Proc object
  Example for registering callbacks with specific events

  handler = Proc.new do |stream|
    stream.on_chunk_event do |event|
      event # => Aws::BedrockRuntime::Types::chunk
    end
    stream.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    stream.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    stream.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    stream.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    stream.on_model_timeout_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelTimeoutException
    end
    stream.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end
  end

  client.invoke_model_with_response_stream( # params input #, event_stream_handler: handler)

Usage pattern c): Hybrid pattern of a) and b)

    handler = Aws::BedrockRuntime::EventStreams::ResponseStream.new
    handler.on_chunk_event do |event|
      event # => Aws::BedrockRuntime::Types::chunk
    end
    handler.on_internal_server_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::internalServerException
    end
    handler.on_model_stream_error_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelStreamErrorException
    end
    handler.on_validation_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::validationException
    end
    handler.on_throttling_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::throttlingException
    end
    handler.on_model_timeout_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::modelTimeoutException
    end
    handler.on_service_unavailable_exception_event do |event|
      event # => Aws::BedrockRuntime::Types::serviceUnavailableException
    end

  client.invoke_model_with_response_stream( # params input #, event_stream_handler: handler) do |stream|
    stream.on_error_event do |event|
      # catch unmodeled error event in the stream
      raise event
      # => Aws::Errors::EventError
      # event.event_type => :error
      # event.error_code => String
      # event.error_message => String
    end
  end

You can also iterate through events after the response complete.

Events are available at resp.body # => Enumerator
For parameter input example, please refer to following request syntax

Request syntax with placeholder values

resp = client.invoke_model_with_response_stream({
  body: "data",
  content_type: "MimeType",
  accept: "MimeType",
  model_id: "InvokeModelIdentifier", # required
  trace: "ENABLED", # accepts ENABLED, DISABLED
  guardrail_identifier: "GuardrailIdentifier",
  guardrail_version: "GuardrailVersion",
  performance_config_latency: "standard", # accepts standard, optimized
})

Response structure

All events are available at resp.body:
resp.body #=> Enumerator
resp.body.event_types #=> [:chunk, :internal_server_exception, :model_stream_error_exception, :validation_exception, :throttling_exception, :model_timeout_exception, :service_unavailable_exception]

For :chunk event available at #on_chunk_event callback and response eventstream enumerator:
event.bytes #=> String

For :internal_server_exception event available at #on_internal_server_exception_event callback and response eventstream enumerator:
event.message #=> String

For :model_stream_error_exception event available at #on_model_stream_error_exception_event callback and response eventstream enumerator:
event.message #=> String
event.original_status_code #=> Integer
event.original_message #=> String

For :validation_exception event available at #on_validation_exception_event callback and response eventstream enumerator:
event.message #=> String

For :throttling_exception event available at #on_throttling_exception_event callback and response eventstream enumerator:
event.message #=> String

For :model_timeout_exception event available at #on_model_timeout_exception_event callback and response eventstream enumerator:
event.message #=> String

For :service_unavailable_exception event available at #on_service_unavailable_exception_event callback and response eventstream enumerator:
event.message #=> String

resp.content_type #=> String
resp.performance_config_latency #=> String, one of "standard", "optimized"

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :body (String, StringIO, File) —

  The prompt and inference parameters in the format specified in the contentType in the header. You must provide the body in JSON format. To see the format and content of the request and response bodies for different models, refer to Inference parameters. For more information, see Run inference in the Bedrock User Guide.

• :content_type (String) —

  The MIME type of the input data in the request. You must specify application/json.

• :accept (String) —

  The desired MIME type of the inference body in the response. The default value is application/json.

• :model_id (required, String) —

  The unique identifier of the model to invoke to run inference.

  The modelId to provide depends on the type of model or throughput that you use:

  • If you use a base model, specify the model ID or its ARN. For a list of model IDs for base models, see Amazon Bedrock base model IDs (on-demand throughput) in the Amazon Bedrock User Guide.

  • If you use an inference profile, specify the inference profile ID or its ARN. For a list of inference profile IDs, see Supported Regions and models for cross-region inference in the Amazon Bedrock User Guide.

  • If you use a provisioned model, specify the ARN of the Provisioned Throughput. For more information, see Run inference using a Provisioned Throughput in the Amazon Bedrock User Guide.

  • If you use a custom model, first purchase Provisioned Throughput for it. Then specify the ARN of the resulting provisioned model. For more information, see Use a custom model in Amazon Bedrock in the Amazon Bedrock User Guide.

  • If you use an imported model, specify the ARN of the imported model. You can get the model ARN from a successful call to CreateModelImportJob or from the Imported models page in the Amazon Bedrock console.

• :trace (String) —

  Specifies whether to enable or disable the Bedrock trace. If enabled, you can see the full Bedrock trace.

• :guardrail_identifier (String) —

  The unique identifier of the guardrail that you want to use. If you don't provide a value, no guardrail is applied to the invocation.

  An error is thrown in the following situations.

  • You don't provide a guardrail identifier but you specify the amazon-bedrock-guardrailConfig field in the request body.

  • You enable the guardrail but the contentType isn't application/json.

  • You provide a guardrail identifier, but guardrailVersion isn't specified.

• :guardrail_version (String) —

  The version number for the guardrail. The value can also be DRAFT.

• :performance_config_latency (String) —

  Model performance settings for the request.

Yields:

• (event_stream_handler)

Returns:

• (Types::InvokeModelWithResponseStreamResponse) —

  Returns a response object which responds to the following methods:

  • #body => Types::ResponseStream
  • #content_type => String
  • #performance_config_latency => String

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 2240

def invoke_model_with_response_stream(params = {}, options = {}, &block)
  params = params.dup
  event_stream_handler = case handler = params.delete(:event_stream_handler)
    when EventStreams::ResponseStream then handler
    when Proc then EventStreams::ResponseStream.new.tap(&handler)
    when nil then EventStreams::ResponseStream.new
    else
      msg = "expected :event_stream_handler to be a block or "\
            "instance of Aws::BedrockRuntime::EventStreams::ResponseStream"\
            ", got `#{handler.inspect}` instead"
      raise ArgumentError, msg
    end

  yield(event_stream_handler) if block_given?

  req = build_request(:invoke_model_with_response_stream, params)

  req.context[:event_stream_handler] = event_stream_handler
  req.handlers.add(Aws::Binary::DecodeHandler, priority: 95)

  req.send_request(options, &block)
end

#list_async_invokes(params = {}) ⇒ Types::ListAsyncInvokesResponse

Lists asynchronous invocations.

The returned response is a pageable response and is Enumerable. For details on usage see PageableResponse.

Examples:

Request syntax with placeholder values

resp = client.list_async_invokes({
  submit_time_after: Time.now,
  submit_time_before: Time.now,
  status_equals: "InProgress", # accepts InProgress, Completed, Failed
  max_results: 1,
  next_token: "PaginationToken",
  sort_by: "SubmissionTime", # accepts SubmissionTime
  sort_order: "Ascending", # accepts Ascending, Descending
})

Response structure

resp.next_token #=> String
resp.async_invoke_summaries #=> Array
resp.async_invoke_summaries[0].invocation_arn #=> String
resp.async_invoke_summaries[0].model_arn #=> String
resp.async_invoke_summaries[0].client_request_token #=> String
resp.async_invoke_summaries[0].status #=> String, one of "InProgress", "Completed", "Failed"
resp.async_invoke_summaries[0].failure_message #=> String
resp.async_invoke_summaries[0].submit_time #=> Time
resp.async_invoke_summaries[0].last_modified_time #=> Time
resp.async_invoke_summaries[0].end_time #=> Time
resp.async_invoke_summaries[0].output_data_config.s3_output_data_config.s3_uri #=> String
resp.async_invoke_summaries[0].output_data_config.s3_output_data_config.kms_key_id #=> String
resp.async_invoke_summaries[0].output_data_config.s3_output_data_config.bucket_owner #=> String

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :submit_time_after (Time, DateTime, Date, Integer, String) —

  Include invocations submitted after this time.

• :submit_time_before (Time, DateTime, Date, Integer, String) —

  Include invocations submitted before this time.

• :status_equals (String) —

  Filter invocations by status.

• :max_results (Integer) —

  The maximum number of invocations to return in one page of results.

• :next_token (String) —

  Specify the pagination token from a previous request to retrieve the next page of results.

• :sort_by (String) —

  How to sort the response.

• :sort_order (String) —

  The sorting order for the response.

Returns:

• (Types::ListAsyncInvokesResponse) —

  Returns a response object which responds to the following methods:

  • #next_token => String
  • #async_invoke_summaries => Array<Types::AsyncInvokeSummary>

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 2326

def list_async_invokes(params = {}, options = {})
  req = build_request(:list_async_invokes, params)
  req.send_request(options)
end

#start_async_invoke(params = {}) ⇒ Types::StartAsyncInvokeResponse

Starts an asynchronous invocation.

This operation requires permission for the bedrock:InvokeModel action.

To deny all inference access to resources that you specify in the modelId field, you need to deny access to the bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream actions. Doing this also denies access to the resource through the Converse API actions (Converse and ConverseStream). For more information see Deny access for inference on specific models.

Examples:

Request syntax with placeholder values

resp = client.start_async_invoke({
  client_request_token: "AsyncInvokeIdempotencyToken",
  model_id: "AsyncInvokeIdentifier", # required
  model_input: { # required
  },
  output_data_config: { # required
    s3_output_data_config: {
      s3_uri: "S3Uri", # required
      kms_key_id: "KmsKeyId",
      bucket_owner: "AccountId",
    },
  },
  tags: [
    {
      key: "TagKey", # required
      value: "TagValue", # required
    },
  ],
})

Response structure

resp.invocation_arn #=> String

Parameters:

• params (Hash) (defaults to: {}) —

  ({})

Options Hash (params):

• :client_request_token (String) —

  Specify idempotency token to ensure that requests are not duplicated.

  A suitable default value is auto-generated. You should normally not need to pass this option.**

• :model_id (required, String) —

  The model to invoke.

• :model_input (required, Hash, Array, String, Numeric, Boolean) —

  Input to send to the model.

  Document type used to carry open content (Hash,Array,String,Numeric,Boolean). A document type value is serialized using the same format as its surroundings and requires no additional encoding or escaping.

• :output_data_config (required, Types::AsyncInvokeOutputDataConfig) —

  Where to store the output.

• :tags (Array<Types::Tag>) —

  Tags to apply to the invocation.

Returns:

• (Types::StartAsyncInvokeResponse) —

  Returns a response object which responds to the following methods:

  • #invocation_arn => String

See Also:

• AWS API Documentation

# File 'gems/aws-sdk-bedrockruntime/lib/aws-sdk-bedrockruntime/client.rb', line 2406

def start_async_invoke(params = {}, options = {})
  req = build_request(:start_async_invoke, params)
  req.send_request(options)
end