OpenTelemetryRealtimeClientBuilderExtensions.UseOpenTelemetry Method

Definition

Namespace:
Microsoft.Extensions.AI
Assembly:
Microsoft.Extensions.AI.dll
Package:
Microsoft.Extensions.AI v10.4.1
Source:
OpenTelemetryRealtimeClientBuilderExtensions.cs

Adds OpenTelemetry support to the realtime client pipeline, following the OpenTelemetry Semantic Conventions for Generative AI systems.

public static Microsoft.Extensions.AI.RealtimeClientBuilder UseOpenTelemetry(this Microsoft.Extensions.AI.RealtimeClientBuilder builder, Microsoft.Extensions.Logging.ILoggerFactory? loggerFactory = default, string? sourceName = default, Action<Microsoft.Extensions.AI.OpenTelemetryRealtimeClient>? configure = default);
static member UseOpenTelemetry : Microsoft.Extensions.AI.RealtimeClientBuilder * Microsoft.Extensions.Logging.ILoggerFactory * string * Action<Microsoft.Extensions.AI.OpenTelemetryRealtimeClient> -> Microsoft.Extensions.AI.RealtimeClientBuilder
<Extension()>
Public Function UseOpenTelemetry (builder As RealtimeClientBuilder, Optional loggerFactory As ILoggerFactory = Nothing, Optional sourceName As String = Nothing, Optional configure As Action(Of OpenTelemetryRealtimeClient) = Nothing) As RealtimeClientBuilder

Parameters

builder
RealtimeClientBuilder

The RealtimeClientBuilder.

loggerFactory
ILoggerFactory

An optional ILoggerFactory to use to create a logger for logging events.

sourceName
String

An optional source name that will be used on the telemetry data.

configure
Action<OpenTelemetryRealtimeClient>

An optional callback that can be used to configure the OpenTelemetryRealtimeClient instance.

Returns

RealtimeClientBuilder

The builder.

Exceptions

ArgumentNullException

builder is null.

Remarks

The draft specification this follows is available at https://opentelemetry.io/docs/specs/semconv/gen-ai/. The specification is still experimental and subject to change; as such, the telemetry output by this client is also subject to change.

The following standard OpenTelemetry GenAI conventions are supported:

  • gen_ai.operation.name - Operation name ("realtime")
  • gen_ai.request.model - Model name from options
  • gen_ai.provider.name - Provider name from metadata
  • gen_ai.response.id - Response ID from ResponseDone messages
  • gen_ai.usage.input_tokens - Input token count
  • gen_ai.usage.output_tokens - Output token count
  • gen_ai.request.max_tokens - Max output tokens from options
  • gen_ai.system_instructions - Instructions from options (sensitive data)
  • gen_ai.conversation.id - Conversation ID from response
  • gen_ai.tool.definitions - Tool definitions (sensitive data)
  • server.address / server.port - Server endpoint info
  • error.type - Error type on failures

Additionally, the following realtime-specific custom attributes are supported:

  • gen_ai.realtime.voice - Voice setting from options
  • gen_ai.realtime.output_modalities - Output modalities (text, audio)
  • gen_ai.realtime.voice_speed - Voice speed setting
  • gen_ai.realtime.session_kind - Session kind (Realtime/Transcription)

Metrics include:

  • gen_ai.client.operation.duration - Duration histogram
  • gen_ai.client.token.usage - Token usage histogram

Applies to