Semantic conventions for OpenAI client operations

Status: Development

[!Warning]

Existing GenAI instrumentations that are using v1.36.0 of this document (or prior):

• SHOULD NOT change the version of the GenAI conventions that they emit by default. Conventions include, but are not limited to, attributes, metric, span and event names, span kind and unit of measure.
• SHOULD introduce an environment variable OTEL_SEMCONV_STABILITY_OPT_IN as a comma-separated list of category-specific values. The list of values includes:
  • gen_ai_latest_experimental - emit the latest experimental version of GenAI conventions (supported by the instrumentation) and do not emit the old one (v1.36.0 or prior).
  • The default behavior is to continue emitting whatever version of the GenAI conventions the instrumentation was emitting (1.36.0 or prior).

This transition plan will be updated to include stable version before the GenAI conventions are marked as stable.

The Semantic Conventions for OpenAI extend and override the Gen AI Semantic Conventions.

Spans

gen_ai.provider.name MUST be set to "openai".

Inference

Status:

Semantic Conventions for OpenAI client spans extend and override the semantic conventions for Gen AI Spans.

gen_ai.provider.name MUST be set to "openai" and SHOULD be provided at span creation time.

Span name SHOULD be {gen_ai.operation.name} {gen_ai.request.model}.

Span kind SHOULD be CLIENT.

Span status SHOULD follow the Recording Errors document.

Attributes:

Key	Stability	Requirement Level	Value Type	Description	Example Values
gen_ai.operation.name		Required	string	The name of the operation being performed. [1]	chat; generate_content; text_completion
gen_ai.request.model		Required	string	The name of the GenAI model a request is being made to. [2]	gpt-4
error.type		Conditionally Required if the operation ended in an error	string	Describes a class of error the operation ended with. [3]	timeout; java.net.UnknownHostException; server_certificate_invalid; 500
gen_ai.conversation.id		Conditionally Required when available	string	The unique identifier for a conversation (session, thread), used to store and correlate messages within this conversation. [4]	conv_5j66UpCpwteGg4YSxUnt7lPY
gen_ai.output.type		Conditionally Required [5]	string	Represents the content type requested by the client. [6]	text; json; image
gen_ai.request.choice.count		Conditionally Required if available, in the request, and !=1	int	The target number of candidate completions to return.	3
gen_ai.request.seed		Conditionally Required if applicable and if the request includes a seed	int	Requests with same seed value more likely to return same result.	100
openai.request.service_tier		Conditionally Required [7]	string	The service tier requested. May be a specific tier, default, or auto.	auto; default
openai.response.service_tier		Conditionally Required [8]	string	The service tier used for the response.	scale; default
server.port		Conditionally Required If server.address is set.	int	GenAI server port. [9]	80; 8080; 443
gen_ai.request.frequency_penalty		Recommended	double	The frequency penalty setting for the GenAI request.	0.1
gen_ai.request.max_tokens		Recommended	int	The maximum number of tokens the model generates for a request.	100
gen_ai.request.presence_penalty		Recommended	double	The presence penalty setting for the GenAI request.	0.1
gen_ai.request.stop_sequences		Recommended	string[]	List of sequences that the model will use to stop generating further tokens.	["forest", "lived"]
gen_ai.request.temperature		Recommended	double	The temperature setting for the GenAI request.	0.0
gen_ai.request.top_p		Recommended	double	The top_p sampling setting for the GenAI request.	1.0
gen_ai.response.finish_reasons		Recommended	string[]	Array of reasons the model stopped generating tokens, corresponding to each generation received.	["stop"]; ["stop", "length"]
gen_ai.response.id		Recommended	string	The unique identifier for the completion.	chatcmpl-123
gen_ai.response.model		Recommended	string	The name of the model that generated the response. [10]	gpt-4-0613
gen_ai.usage.input_tokens		Recommended	int	The number of tokens used in the GenAI input (prompt).	100
gen_ai.usage.output_tokens		Recommended	int	The number of tokens used in the GenAI response (completion).	180
openai.response.system_fingerprint		Recommended	string	A fingerprint to track any eventual change in the Generative AI environment.	fp_44709d6fcb
server.address		Recommended	string	GenAI server address. [11]	example.com; 10.1.2.80; /tmp/my.sock
gen_ai.input.messages		Opt-In	any	The chat history provided to the model as an input. [12]	[   {     “role”: “user”,     “parts”: [       {         “type”: “text”,         “content”: “Weather in Paris?"       }     ]   },   {     “role”: “assistant”,     “parts”: [       {         “type”: “tool_call”,         “id”: “call_VSPygqKTWdrhaFErNvMV18Yl”,         “name”: “get_weather”,         “arguments”: {           “location”: “Paris”         }       }     ]   },   {     “role”: “tool”,     “parts”: [       {         “type”: “tool_call_response”,         “id”: " call_VSPygqKTWdrhaFErNvMV18Yl”,         “result”: “rainy, 57°F”       }     ]   } ]
gen_ai.output.messages		Opt-In	any	Messages returned by the model where each message represents a specific model response (choice, candidate). [13]	[   {     “role”: “assistant”,     “parts”: [       {         “type”: “text”,         “content”: “The weather in Paris is currently rainy with a temperature of 57°F."       }     ],     “finish_reason”: “stop”   } ]
gen_ai.system_instructions		Opt-In	any	The system message or instructions provided to the GenAI model separately from the chat history. [14]	[   {     “type”: “text”,     “content”: “You are an Agent that greet users, always use greetings tool to respond”   } ]; [   {     “type”: “text”,     “content”: “You are a language translator."   },   {     “type”: “text”,     “content”: “Your mission is to translate text in English to French."   } ]
gen_ai.tool.definitions		Opt-In	any	The list of source system tool definitions available to the GenAI agent or model. [15]	[   {     “type”: “function”,     “name”: “get_current_weather”,     “description”: “Get the current weather in a given location”,     “parameters”: {       “type”: “object”,       “properties”: {         “location”: {           “type”: “string”,           “description”: “The city and state, e.g. San Francisco, CA”         },         “unit”: {           “type”: “string”,           “enum”: [             “celsius”,             “fahrenheit”           ]         }       },       “required”: [         “location”,         “unit”       ]     }   } ]

[1] gen_ai.operation.name: If one of the predefined values applies, but specific system uses a different name it’s RECOMMENDED to document it in the semantic conventions for specific GenAI system and use system-specific name in the instrumentation. If a different name is not documented, instrumentation libraries SHOULD use applicable predefined value.

[2] gen_ai.request.model: The name of the GenAI model a request is being made to. If the model is supplied by a vendor, then the value must be the exact name of the model requested. If the model is a fine-tuned custom model, the value should have a more specific name than the base model that’s been fine-tuned.

[3] error.type: The error.type SHOULD match the error code returned by the Generative AI provider or the client library, the canonical name of exception that occurred, or another low-cardinality error identifier. Instrumentations SHOULD document the list of errors they report.

[4] gen_ai.conversation.id: Instrumentations SHOULD populate conversation id when they have it readily available for a given operation, for example:

• when client framework being instrumented manages conversation history (see LlamaIndex chat store)

• when instrumenting GenAI client libraries that maintain conversation on the backend side (see AWS Bedrock agent sessions, OpenAI Assistant threads)

Application developers that manage conversation history MAY add conversation id to GenAI and other spans or logs using custom span or log record processors or hooks provided by instrumentation libraries.

[5] gen_ai.output.type: when applicable and if the request includes an output format.

[6] gen_ai.output.type: This attribute SHOULD be set to the output type requested by the client:

• json for structured outputs with defined or undefined schema
• image for image output
• speech for speech output
• text for plain text output

The attribute specifies the output modality and not the actual output format. For example, if an image is requested, the actual output could be a URL pointing to an image file.

Additional output format details may be recorded in the future in the gen_ai.output.{type}.* attributes.

[7] openai.request.service_tier: if the request includes a service_tier and the value is not ‘auto’

[8] openai.response.service_tier: if the response was received and includes a service_tier

[9] server.port: When observed from the client side, and when communicating through an intermediary, server.port SHOULD represent the server port behind any intermediaries, for example proxies, if it’s available.

[10] gen_ai.response.model: If available. The name of the GenAI model that provided the response. If the model is supplied by a vendor, then the value must be the exact name of the model actually used. If the model is a fine-tuned custom model, the value should have a more specific name than the base model that’s been fine-tuned.

[11] server.address: When observed from the client side, and when communicating through an intermediary, server.address SHOULD represent the server address behind any intermediaries, for example proxies, if it’s available.

[12] gen_ai.input.messages: Instrumentations MUST follow Input messages JSON schema. When the attribute is recorded on events, it MUST be recorded in structured form. When recorded on spans, it MAY be recorded as a JSON string if structured format is not supported and SHOULD be recorded in structured form otherwise.

Messages MUST be provided in the order they were sent to the model. Instrumentations MAY provide a way for users to filter or truncate input messages.

[!Warning] This attribute is likely to contain sensitive information including user/PII data.

See Recording content on attributes section for more details.

[13] gen_ai.output.messages: Instrumentations MUST follow Output messages JSON schema

Each message represents a single output choice/candidate generated by the model. Each message corresponds to exactly one generation (choice/candidate) and vice versa - one choice cannot be split across multiple messages or one message cannot contain parts from multiple choices.

When the attribute is recorded on events, it MUST be recorded in structured form. When recorded on spans, it MAY be recorded as a JSON string if structured format is not supported and SHOULD be recorded in structured form otherwise.

Instrumentations MAY provide a way for users to filter or truncate output messages.

[!Warning] This attribute is likely to contain sensitive information including user/PII data.

See Recording content on attributes section for more details.

[14] gen_ai.system_instructions: This attribute SHOULD be used when the corresponding provider or API allows to provide system instructions or messages separately from the chat history.

Instructions that are part of the chat history SHOULD be recorded in gen_ai.input.messages attribute instead.

Instrumentations MUST follow System instructions JSON schema.

When recorded on spans, it MAY be recorded as a JSON string if structured format is not supported and SHOULD be recorded in structured form otherwise.

Instrumentations MAY provide a way for users to filter or truncate system instructions.

[!Warning] This attribute may contain sensitive information.

See Recording content on attributes section for more details.

[15] gen_ai.tool.definitions: The value of this attribute matches source system tool definition format.

It’s expected to be an array of objects where each object represents a tool definition. In case a serialized string is available to the instrumentation, the instrumentation SHOULD do the best effort to deserialize it to an array. When recorded on spans, it MAY be recorded as a JSON string if structured format is not supported and SHOULD be recorded in structured form otherwise.

Since this attribute could be large, it’s NOT RECOMMENDED to populate it by default. Instrumentations MAY provide a way to enable populating this attribute.

error.type has the following list of well-known values. If one of them applies, then the respective value MUST be used; otherwise, a custom value MAY be used.

Value	Description	Stability
_OTHER	A fallback error value to be used when the instrumentation doesn’t define a custom value.

gen_ai.operation.name has the following list of well-known values. If one of them applies, then the respective value MUST be used; otherwise, a custom value MAY be used.

Value	Description	Stability
chat	Chat completion operation such as OpenAI Chat API
create_agent	Create GenAI agent
embeddings	Embeddings operation such as OpenAI Create embeddings API
execute_tool	Execute a tool
generate_content	Multimodal content generation operation such as Gemini Generate Content
invoke_agent	Invoke GenAI agent
text_completion	Text completions operation such as OpenAI Completions API (Legacy)

gen_ai.output.type has the following list of well-known values. If one of them applies, then the respective value MUST be used; otherwise, a custom value MAY be used.

Value	Description	Stability
image	Image
json	JSON object with known or unknown schema
speech	Speech
text	Plain text

openai.request.service_tier has the following list of well-known values. If one of them applies, then the respective value MUST be used; otherwise, a custom value MAY be used.

Value	Description	Stability
auto	The system will utilize scale tier credits until they are exhausted.
default	The system will utilize the default scale tier.

Embeddings

See common embeddings span definition.

Metrics

OpenAI metrics follow Generative AI metrics with the noted additional attributes. Individual systems may include additional system-specific attributes. It is recommended to check system-specific documentation, if available.

Metric: gen_ai.client.token.usage

Reports the usage of tokens following the common gen_ai.client.token.usage definition.

Additional attributes:

Attributes:

Key	Stability	Requirement Level	Value Type	Description	Example Values
openai.response.service_tier		Recommended	string	The service tier used for the response.	scale; default
openai.response.system_fingerprint		Recommended	string	A fingerprint to track any eventual change in the Generative AI environment.	fp_44709d6fcb

Metric: gen_ai.client.operation.duration

Measures the to complete an operation following the common gen_ai.client.operation.duration definition.

Additional attributes:

Attributes:

Key	Stability	Requirement Level	Value Type	Description	Example Values
openai.response.service_tier		Recommended	string	The service tier used for the response.	scale; default
openai.response.system_fingerprint		Recommended	string	A fingerprint to track any eventual change in the Generative AI environment.	fp_44709d6fcb