View Source OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics (OpenTelemetry.SemConv v1.27.0)

OpenTelemetry Semantic Conventions for Gen_Ai metrics.

Summary

GenAI operation duration

Measures number of input and output tokens used

Generative AI server request duration such as time-to-last byte or last output token

Time per output token generated after the first token for successful responses

Time to generate first token for successful responses

@spec gen_ai_client_operation_duration() :: :"gen_ai.client.operation.duration"

GenAI operation duration

Instrument: histogram
Unit: s

iex> OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics.gen_ai_client_operation_duration()
:"gen_ai.client.operation.duration"

?GEN_AI_CLIENT_OPERATION_DURATION.
'gen_ai.client.operation.duration'

@spec gen_ai_client_token_usage() :: :"gen_ai.client.token.usage"

Measures number of input and output tokens used

Instrument: histogram
Unit: {token}

iex> OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics.gen_ai_client_token_usage()
:"gen_ai.client.token.usage"

?GEN_AI_CLIENT_TOKEN_USAGE.
'gen_ai.client.token.usage'

@spec gen_ai_server_request_duration() :: :"gen_ai.server.request.duration"

Generative AI server request duration such as time-to-last byte or last output token

Instrument: histogram
Unit: s

iex> OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics.gen_ai_server_request_duration()
:"gen_ai.server.request.duration"

?GEN_AI_SERVER_REQUEST_DURATION.
'gen_ai.server.request.duration'

@spec gen_ai_server_time_per_output_token() :: :"gen_ai.server.time_per_output_token"

Time per output token generated after the first token for successful responses

Instrument: histogram
Unit: s

iex> OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics.gen_ai_server_time_per_output_token()
:"gen_ai.server.time_per_output_token"

?GEN_AI_SERVER_TIME_PER_OUTPUT_TOKEN.
'gen_ai.server.time_per_output_token'

@spec gen_ai_server_time_to_first_token() :: :"gen_ai.server.time_to_first_token"

Time to generate first token for successful responses

Instrument: histogram
Unit: s

iex> OpenTelemetry.SemConv.Incubating.Metrics.GenAiMetrics.gen_ai_server_time_to_first_token()
:"gen_ai.server.time_to_first_token"

?GEN_AI_SERVER_TIME_TO_FIRST_TOKEN.
'gen_ai.server.time_to_first_token'