vellum-ai 0.3.20__tar.gz → 0.3.21__tar.gz

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vellum-ai
-Version: 0.3.20
+Version: 0.3.21
 Summary: 
 Requires-Python: >=3.8,<4.0
 Classifier: Programming Language :: Python :: 3

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "vellum-ai"
-version = "0.3.20"
+version = "0.3.21"
 description = ""
 readme = "README.md"
 authors = []

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/__init__.py

@@ -148,6 +148,7 @@ from .types import (
     MetadataFilterConfigRequest,
     MetadataFilterRuleCombinator,
     MetadataFilterRuleRequest,
+    MlModelUsage,
     ModelVersionBuildConfig,
     ModelVersionExecConfig,
     ModelVersionExecConfigParameters,
@@ -641,6 +642,7 @@ __all__ = [
     "MetadataFilterConfigRequest",
     "MetadataFilterRuleCombinator",
     "MetadataFilterRuleRequest",
+    "MlModelUsage",
     "ModelVersionBuildConfig",
     "ModelVersionExecConfig",
     "ModelVersionExecConfigParameters",

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/client.py

@@ -167,6 +167,7 @@ class Vellum:
                 deployment_release_tag=True,
                 prompt_version_id=True,
                 finish_reason=True,
+                usage=True,
             ),
             raw_overrides=RawPromptExecutionOverridesRequest(
                 body={"string": {"key": "value"}},
@@ -301,6 +302,7 @@ class Vellum:
                 deployment_release_tag=True,
                 prompt_version_id=True,
                 finish_reason=True,
+                usage=True,
             ),
             raw_overrides=RawPromptExecutionOverridesRequest(
                 body={"string": {"key": "value"}},
@@ -1123,6 +1125,7 @@ class AsyncVellum:
                 deployment_release_tag=True,
                 prompt_version_id=True,
                 finish_reason=True,
+                usage=True,
             ),
             raw_overrides=RawPromptExecutionOverridesRequest(
                 body={"string": {"key": "value"}},
@@ -1257,6 +1260,7 @@ class AsyncVellum:
                 deployment_release_tag=True,
                 prompt_version_id=True,
                 finish_reason=True,
+                usage=True,
             ),
             raw_overrides=RawPromptExecutionOverridesRequest(
                 body={"string": {"key": "value"}},

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/core/client_wrapper.py

@@ -18,7 +18,7 @@ class BaseClientWrapper:
         headers: typing.Dict[str, str] = {
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "vellum-ai",
-            "X-Fern-SDK-Version": "0.3.20",
+            "X-Fern-SDK-Version": "0.3.21",
         }
         headers["X_API_KEY"] = self.api_key
         return headers

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/types/__init__.py

@@ -165,6 +165,7 @@ from .logprobs_enum import LogprobsEnum
 from .metadata_filter_config_request import MetadataFilterConfigRequest
 from .metadata_filter_rule_combinator import MetadataFilterRuleCombinator
 from .metadata_filter_rule_request import MetadataFilterRuleRequest
+from .ml_model_usage import MlModelUsage
 from .model_version_build_config import ModelVersionBuildConfig
 from .model_version_exec_config import ModelVersionExecConfig
 from .model_version_exec_config_parameters import ModelVersionExecConfigParameters
@@ -673,6 +674,7 @@ __all__ = [
     "MetadataFilterConfigRequest",
     "MetadataFilterRuleCombinator",
     "MetadataFilterRuleRequest",
+    "MlModelUsage",
     "ModelVersionBuildConfig",
     "ModelVersionExecConfig",
     "ModelVersionExecConfigParameters",

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/types/fulfilled_prompt_execution_meta.py

@@ -5,6 +5,7 @@ import typing
 
 from ..core.datetime_utils import serialize_datetime
 from .finish_reason_enum import FinishReasonEnum
+from .ml_model_usage import MlModelUsage
 
 try:
     import pydantic.v1 as pydantic  # type: ignore
@@ -19,6 +20,7 @@ class FulfilledPromptExecutionMeta(pydantic.BaseModel):
 
     latency: typing.Optional[int] = None
     finish_reason: typing.Optional[FinishReasonEnum] = None
+    usage: typing.Optional[MlModelUsage] = None
 
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}

vellum_ai-0.3.21/src/vellum/types/ml_model_usage.py

@@ -0,0 +1,33 @@
+# This file was auto-generated by Fern from our API Definition.
+
+import datetime as dt
+import typing
+
+from ..core.datetime_utils import serialize_datetime
+
+try:
+    import pydantic.v1 as pydantic  # type: ignore
+except ImportError:
+    import pydantic  # type: ignore
+
+
+class MlModelUsage(pydantic.BaseModel):
+    output_token_count: typing.Optional[int] = None
+    input_token_count: typing.Optional[int] = None
+    input_char_count: typing.Optional[int] = None
+    output_char_count: typing.Optional[int] = None
+    compute_nanos: typing.Optional[int] = None
+
+    def json(self, **kwargs: typing.Any) -> str:
+        kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}
+        return super().json(**kwargs_with_defaults)
+
+    def dict(self, **kwargs: typing.Any) -> typing.Dict[str, typing.Any]:
+        kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}
+        return super().dict(**kwargs_with_defaults)
+
+    class Config:
+        frozen = True
+        smart_union = True
+        extra = pydantic.Extra.allow
+        json_encoders = {dt.datetime: serialize_datetime}

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/types/prompt_deployment_expand_meta_request_request.py

@@ -37,6 +37,11 @@ class PromptDeploymentExpandMetaRequestRequest(pydantic.BaseModel):
     If enabled, the response will include the reason provided by the model for why the execution finished.
     """
 
+    usage: typing.Optional[bool] = pydantic.Field(default=None)
+    """
+    If enabled, the response will include model host usage tracking. This may increase latency for some model hosts.
+    """
+
     def json(self, **kwargs: typing.Any) -> str:
         kwargs_with_defaults: typing.Any = {"by_alias": True, "exclude_unset": True, **kwargs}
         return super().json(**kwargs_with_defaults)

{vellum_ai-0.3.20 → vellum_ai-0.3.21}/src/vellum/types/prompt_execution_meta.py

@@ -5,6 +5,7 @@ import typing
 
 from ..core.datetime_utils import serialize_datetime
 from .finish_reason_enum import FinishReasonEnum
+from .ml_model_usage import MlModelUsage
 
 try:
     import pydantic.v1 as pydantic  # type: ignore
@@ -17,6 +18,7 @@ class PromptExecutionMeta(pydantic.BaseModel):
     The subset of the metadata tracked by Vellum during prompt execution that the request opted into with `expand_meta`.
     """
 
+    usage: typing.Optional[MlModelUsage] = None
     model_name: typing.Optional[str] = None
     latency: typing.Optional[int] = None
     deployment_release_tag: typing.Optional[str] = None