Module `livekit.agents.metrics`

Functions

def log_metrics(metrics: AgentMetrics, *, logger: logging.Logger | None = None) ‑> None

Expand source code

def log_metrics(metrics: AgentMetrics, *, logger: logging.Logger | None = None) -> None:
    if logger is None:
        logger = default_logger

    if isinstance(metrics, LLMMetrics):
        logger.info(
            "LLM metrics",
            extra={
                "ttft": round(metrics.ttft, 2),
                "prompt_tokens": metrics.prompt_tokens,
                "prompt_cached_tokens": metrics.prompt_cached_tokens,
                "completion_tokens": metrics.completion_tokens,
                "tokens_per_second": round(metrics.tokens_per_second, 2),
            },
        )
    elif isinstance(metrics, RealtimeModelMetrics):
        logger.info(
            "RealtimeModel metrics",
            extra={
                "ttft": round(metrics.ttft, 2),
                "input_tokens": metrics.input_tokens,
                "cached_input_tokens": metrics.input_token_details.cached_tokens,
                "output_tokens": metrics.output_tokens,
                "total_tokens": metrics.total_tokens,
                "tokens_per_second": round(metrics.tokens_per_second, 2),
            },
        )
    elif isinstance(metrics, TTSMetrics):
        logger.info(
            "TTS metrics",
            extra={
                "ttfb": metrics.ttfb,
                "audio_duration": round(metrics.audio_duration, 2),
            },
        )
    elif isinstance(metrics, EOUMetrics):
        logger.info(
            "EOU metrics",
            extra={
                "end_of_utterance_delay": round(metrics.end_of_utterance_delay, 2),
                "transcription_delay": round(metrics.transcription_delay, 2),
            },
        )
    elif isinstance(metrics, STTMetrics):
        logger.info(
            "STT metrics",
            extra={
                "audio_duration": round(metrics.audio_duration, 2),
            },
        )

Classes

class EOUMetrics (**data: Any)

Expand source code

class EOUMetrics(BaseModel):
    type: Literal["eou_metrics"] = "eou_metrics"
    timestamp: float
    end_of_utterance_delay: float
    """Amount of time between the end of speech from VAD and the decision to end the user's turn.
    Set to 0.0 if the end of speech was not detected.
    """

    transcription_delay: float
    """Time taken to obtain the transcript after the end of the user's speech.
    Set to 0.0 if the end of speech was not detected.
    """

    on_user_turn_completed_delay: float
    """Time taken to invoke the user's `Agent.on_user_turn_completed` callback."""

    last_speaking_time: float
    """The time the user stopped speaking."""

    speech_id: str | None = None

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var end_of_utterance_delay : float: Amount of time between the end of speech from VAD and the decision to end the user's turn. Set to 0.0 if the end of speech was not detected.
var last_speaking_time : float: The time the user stopped speaking.
var model_config
var on_user_turn_completed_delay : float: Time taken to invoke the user's Agent.on_user_turn_completed callback.
var speech_id : str | None
var timestamp : float
var transcription_delay : float: Time taken to obtain the transcript after the end of the user's speech. Set to 0.0 if the end of speech was not detected.
var type : Literal['eou_metrics']

class LLMMetrics (**data: Any)

Expand source code

class LLMMetrics(BaseModel):
    type: Literal["llm_metrics"] = "llm_metrics"
    label: str
    request_id: str
    timestamp: float
    duration: float
    ttft: float
    cancelled: bool
    completion_tokens: int
    prompt_tokens: int
    prompt_cached_tokens: int
    total_tokens: int
    tokens_per_second: float
    speech_id: str | None = None

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var cancelled : bool
var completion_tokens : int
var duration : float
var label : str
var model_config
var prompt_cached_tokens : int
var prompt_tokens : int
var request_id : str
var speech_id : str | None
var timestamp : float
var tokens_per_second : float
var total_tokens : int
var ttft : float
var type : Literal['llm_metrics']

class RealtimeModelMetrics (**data: Any)

Expand source code

class RealtimeModelMetrics(BaseModel):
    class CachedTokenDetails(BaseModel):
        audio_tokens: int
        text_tokens: int
        image_tokens: int

    class InputTokenDetails(BaseModel):
        audio_tokens: int
        text_tokens: int
        image_tokens: int
        cached_tokens: int
        cached_tokens_details: RealtimeModelMetrics.CachedTokenDetails | None

    class OutputTokenDetails(BaseModel):
        text_tokens: int
        audio_tokens: int
        image_tokens: int

    type: Literal["realtime_model_metrics"] = "realtime_model_metrics"
    label: str
    model: str
    request_id: str
    timestamp: float
    """The timestamp of the response creation."""
    duration: float
    """The duration of the response from created to done in seconds."""
    ttft: float
    """Time to first audio token in seconds. -1 if no audio token was sent."""
    cancelled: bool
    """Whether the request was cancelled."""
    input_tokens: int
    """The number of input tokens used in the Response, including text and audio tokens."""
    output_tokens: int
    """The number of output tokens sent in the Response, including text and audio tokens."""
    total_tokens: int
    """The total number of tokens in the Response."""
    tokens_per_second: float
    """The number of tokens per second."""
    input_token_details: InputTokenDetails
    """Details about the input tokens used in the Response."""
    output_token_details: OutputTokenDetails
    """Details about the output tokens used in the Response."""

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var CachedTokenDetails

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

var InputTokenDetails

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

var OutputTokenDetails

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

var cancelled : bool

Whether the request was cancelled.

var duration : float

The duration of the response from created to done in seconds.

var input_token_details : livekit.agents.metrics.base.RealtimeModelMetrics.InputTokenDetails

Details about the input tokens used in the Response.

var input_tokens : int

The number of input tokens used in the Response, including text and audio tokens.

var label : str

var model : str

var model_config

var output_token_details : livekit.agents.metrics.base.RealtimeModelMetrics.OutputTokenDetails

Details about the output tokens used in the Response.

var output_tokens : int

The number of output tokens sent in the Response, including text and audio tokens.

var request_id : str

var timestamp : float

The timestamp of the response creation.

var tokens_per_second : float

The number of tokens per second.

var total_tokens : int

The total number of tokens in the Response.

var ttft : float

Time to first audio token in seconds. -1 if no audio token was sent.

var type : Literal['realtime_model_metrics']

class STTMetrics (**data: Any)

Expand source code

class STTMetrics(BaseModel):
    type: Literal["stt_metrics"] = "stt_metrics"
    label: str
    request_id: str
    timestamp: float
    duration: float
    """The request duration in seconds, 0.0 if the STT is streaming."""
    audio_duration: float
    """The duration of the pushed audio in seconds."""
    streamed: bool
    """Whether the STT is streaming (e.g using websocket)."""

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var audio_duration : float: The duration of the pushed audio in seconds.
var duration : float: The request duration in seconds, 0.0 if the STT is streaming.
var label : str
var model_config
var request_id : str
var streamed : bool: Whether the STT is streaming (e.g using websocket).
var timestamp : float
var type : Literal['stt_metrics']

class TTSMetrics (**data: Any)

Expand source code

class TTSMetrics(BaseModel):
    type: Literal["tts_metrics"] = "tts_metrics"
    label: str
    request_id: str
    timestamp: float
    ttfb: float
    duration: float
    audio_duration: float
    cancelled: bool
    characters_count: int
    streamed: bool
    segment_id: str | None = None
    speech_id: str | None = None

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var audio_duration : float
var cancelled : bool
var characters_count : int
var duration : float
var label : str
var model_config
var request_id : str
var segment_id : str | None
var speech_id : str | None
var streamed : bool
var timestamp : float
var ttfb : float
var type : Literal['tts_metrics']

class UsageCollector

Expand source code

class UsageCollector:
    def __init__(self) -> None:
        self._summary = UsageSummary()

    def __call__(self, metrics: AgentMetrics) -> None:
        self.collect(metrics)

    def collect(self, metrics: AgentMetrics) -> None:
        if isinstance(metrics, LLMMetrics):
            self._summary.llm_prompt_tokens += metrics.prompt_tokens
            self._summary.llm_prompt_cached_tokens += metrics.prompt_cached_tokens
            self._summary.llm_completion_tokens += metrics.completion_tokens

        elif isinstance(metrics, RealtimeModelMetrics):
            self._summary.llm_input_audio_tokens += metrics.input_token_details.audio_tokens
            self._summary.llm_output_audio_tokens += metrics.output_token_details.audio_tokens
            self._summary.llm_prompt_tokens += metrics.input_tokens
            self._summary.llm_prompt_cached_tokens += metrics.input_token_details.cached_tokens
            self._summary.llm_completion_tokens += metrics.output_tokens

        elif isinstance(metrics, TTSMetrics):
            self._summary.tts_characters_count += metrics.characters_count
            self._summary.tts_audio_duration += metrics.audio_duration

        elif isinstance(metrics, STTMetrics):
            self._summary.stt_audio_duration += metrics.audio_duration

    def get_summary(self) -> UsageSummary:
        return deepcopy(self._summary)

Methods

def collect(self, metrics: livekit.agents.metrics.base.STTMetrics | livekit.agents.metrics.base.LLMMetrics | livekit.agents.metrics.base.TTSMetrics | livekit.agents.metrics.base.VADMetrics | livekit.agents.metrics.base.EOUMetrics | livekit.agents.metrics.base.RealtimeModelMetrics) ‑> None

Expand source code

def collect(self, metrics: AgentMetrics) -> None:
    if isinstance(metrics, LLMMetrics):
        self._summary.llm_prompt_tokens += metrics.prompt_tokens
        self._summary.llm_prompt_cached_tokens += metrics.prompt_cached_tokens
        self._summary.llm_completion_tokens += metrics.completion_tokens

    elif isinstance(metrics, RealtimeModelMetrics):
        self._summary.llm_input_audio_tokens += metrics.input_token_details.audio_tokens
        self._summary.llm_output_audio_tokens += metrics.output_token_details.audio_tokens
        self._summary.llm_prompt_tokens += metrics.input_tokens
        self._summary.llm_prompt_cached_tokens += metrics.input_token_details.cached_tokens
        self._summary.llm_completion_tokens += metrics.output_tokens

    elif isinstance(metrics, TTSMetrics):
        self._summary.tts_characters_count += metrics.characters_count
        self._summary.tts_audio_duration += metrics.audio_duration

    elif isinstance(metrics, STTMetrics):
        self._summary.stt_audio_duration += metrics.audio_duration

def get_summary(self) ‑> livekit.agents.metrics.usage_collector.UsageSummary

Expand source code

def get_summary(self) -> UsageSummary:
    return deepcopy(self._summary)

class UsageSummary (llm_prompt_tokens: int = 0, llm_prompt_cached_tokens: int = 0, llm_input_audio_tokens: int = 0, llm_completion_tokens: int = 0, llm_output_audio_tokens: int = 0, tts_characters_count: int = 0, tts_audio_duration: float = 0.0, stt_audio_duration: float = 0.0)

Expand source code

@dataclass
class UsageSummary:
    llm_prompt_tokens: int = 0
    llm_prompt_cached_tokens: int = 0
    llm_input_audio_tokens: int = 0
    llm_completion_tokens: int = 0
    llm_output_audio_tokens: int = 0
    tts_characters_count: int = 0
    tts_audio_duration: float = 0.0
    stt_audio_duration: float = 0.0

UsageSummary(llm_prompt_tokens: int = 0, llm_prompt_cached_tokens: int = 0, llm_input_audio_tokens: int = 0, llm_completion_tokens: int = 0, llm_output_audio_tokens: int = 0, tts_characters_count: int = 0, tts_audio_duration: float = 0.0, stt_audio_duration: float = 0.0)

Instance variables

var llm_completion_tokens : int
var llm_input_audio_tokens : int
var llm_output_audio_tokens : int
var llm_prompt_cached_tokens : int
var llm_prompt_tokens : int
var stt_audio_duration : float
var tts_audio_duration : float
var tts_characters_count : int

class VADMetrics (**data: Any)

Expand source code

class VADMetrics(BaseModel):
    type: Literal["vad_metrics"] = "vad_metrics"
    label: str
    timestamp: float
    idle_time: float
    inference_duration_total: float
    inference_count: int

Usage docs: https://docs.pydantic.dev/2.10/concepts/models/

A base class for creating Pydantic models.

Attributes

__class_vars__: The names of the class variables defined on the model.
__private_attributes__: Metadata about the private attributes of the model.
__signature__: The synthesized __init__ [Signature][inspect.Signature] of the model.
__pydantic_complete__: Whether model building is completed, or if there are still undefined fields.
__pydantic_core_schema__: The core schema of the model.
__pydantic_custom_init__: Whether the model has a custom __init__ function.
__pydantic_decorators__: Metadata containing the decorators defined on the model. This replaces Model.__validators__ and Model.__root_validators__ from Pydantic V1.
__pydantic_generic_metadata__: Metadata for generic models; contains data used for a similar purpose to args, origin, parameters in typing-module generics. May eventually be replaced by these.
__pydantic_parent_namespace__: Parent namespace of the model, used for automatic rebuilding of models.
__pydantic_post_init__: The name of the post-init method for the model, if defined.
__pydantic_root_model__: Whether the model is a [RootModel][pydantic.root_model.RootModel].
__pydantic_serializer__: The pydantic-core SchemaSerializer used to dump instances of the model.
__pydantic_validator__: The pydantic-core SchemaValidator used to validate instances of the model.
__pydantic_fields__: A dictionary of field names and their corresponding [FieldInfo][pydantic.fields.FieldInfo] objects.
__pydantic_computed_fields__: A dictionary of computed field names and their corresponding [ComputedFieldInfo][pydantic.fields.ComputedFieldInfo] objects.
__pydantic_extra__: A dictionary containing extra values, if [extra][pydantic.config.ConfigDict.extra] is set to 'allow'.
__pydantic_fields_set__: The names of fields explicitly set during instantiation.
__pydantic_private__: Values of private attributes set on the model instance.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Ancestors

pydantic.main.BaseModel

Class variables

var idle_time : float
var inference_count : int
var inference_duration_total : float
var label : str
var model_config
var timestamp : float
var type : Literal['vad_metrics']