mirascope 1.18.3__py3-none-any.whl → 1.18.4__py3-none-any.whl

mirascope/core/gemini/call_response_chunk.py

@@ -78,6 +78,11 @@ class GeminiCallResponseChunk(
         """Returns the number of input tokens."""
         return None
 
+    @property
+    def cached_tokens(self) -> None:
+        """Returns the number of cached tokens."""
+        return None
+
     @property
     def output_tokens(self) -> None:
         """Returns the number of output tokens."""

mirascope/core/gemini/stream.py

@@ -69,7 +69,9 @@ class GeminiStream(
     @property
     def cost(self) -> float | None:
         """Returns the cost of the call."""
-        return calculate_cost(self.input_tokens, self.output_tokens, self.model)
+        return calculate_cost(
+            self.input_tokens, self.cached_tokens, self.output_tokens, self.model
+        )
 
     def _construct_message_param(
         self, tool_calls: list[FunctionCall] | None = None, content: str | None = None

mirascope/core/google/__init__.py

@@ -17,13 +17,13 @@ from .tool import GoogleTool
 GoogleMessageParam: TypeAlias = ContentDict | FunctionResponse | BaseMessageParam
 
 __all__ = [
-    "call",
-    "GoogleDynamicConfig",
     "GoogleCallParams",
     "GoogleCallResponse",
     "GoogleCallResponseChunk",
+    "GoogleDynamicConfig",
     "GoogleMessageParam",
     "GoogleStream",
     "GoogleTool",
+    "call",
     "google_call",
 ]

mirascope/core/google/_utils/_calculate_cost.py

@@ -2,7 +2,10 @@
 
 
 def calculate_cost(
-    input_tokens: int | float | None, output_tokens: int | float | None, model: str
+    input_tokens: int | float | None,
+    cached_tokens: int | float | None,
+    output_tokens: int | float | None,
+    model: str,
 ) -> float | None:
     """Calculate the cost of a Google API call.
 
@@ -10,16 +13,31 @@ def calculate_cost(
 
     Pricing (per 1M tokens):
 
-    Model                   Input (<128K)    Output (<128K)    Input (>128K)    Output (>128K)
-    gemini-2.0-flash       $0.10            $0.40            $0.10            $0.40
-    gemini-2.0-flash-lite  $0.075           $0.30            $0.075           $0.30
-    gemini-1.5-flash       $0.075           $0.30            $0.15            $0.60
-    gemini-1.5-flash-8b    $0.0375          $0.15            $0.075           $0.30
-    gemini-1.5-pro         $1.25            $5.00            $2.50            $10.00
-    gemini-1.0-pro         $0.50            $1.50            $0.50            $1.50
+    Model                                 Input (<128K)    Output (<128K)   Input (>128K)    Output (>128K)   Cached
+    gemini-2.0-pro                        $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-2.0-pro-preview-1206           $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-2.0-flash                      $0.10            $0.40            $0.10            $0.40            $0.0375
+    gemini-2.0-flash-latest               $0.10            $0.40            $0.10            $0.40            $0.0375
+    gemini-2.0-flash-001                  $0.10            $0.40            $0.10            $0.40            $0.0375
+    gemini-2.0-flash-lite                 $0.075           $0.30            $0.075           $0.30            $0.0375
+    gemini-2.0-flash-lite-preview-02-05   $0.075           $0.30            $0.075           $0.30            $0.0375
+    gemini-1.5-pro                        $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-1.5-pro-latest                 $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-1.5-pro-001                    $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-1.5-pro-002                    $1.25            $5.00            $2.50            $10.00           $0.625
+    gemini-1.5-flash                      $0.075           $0.30            $0.15            $0.60            $0.0375
+    gemini-1.5-flash-latest               $0.075           $0.30            $0.15            $0.60            $0.0375
+    gemini-1.5-flash-001                  $0.075           $0.30            $0.15            $0.60            $0.0375
+    gemini-1.5-flash-002                  $0.075           $0.30            $0.15            $0.60            $0.0375
+    gemini-1.5-flash-8b                   $0.0375          $0.15            $0.075           $0.30            $0.025
+    gemini-1.5-flash-8b-latest            $0.0375          $0.15            $0.075           $0.30            $0.025
+    gemini-1.5-flash-8b-001               $0.0375          $0.15            $0.075           $0.30            $0.025
+    gemini-1.5-flash-8b-002               $0.0375          $0.15            $0.075           $0.30            $0.025
+    gemini-1.0-pro                        $0.50            $1.50            $0.50            $1.50            $0.00
 
     Args:
         input_tokens: Number of input tokens
+        cached_tokens: Number of cached tokens
         output_tokens: Number of output tokens
         model: Model name to use for pricing calculation
 
@@ -27,47 +45,154 @@ def calculate_cost(
         Total cost in USD or None if invalid input
     """
     pricing = {
+        "gemini-2.0-pro": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
+        },
+        "gemini-2.0-pro-preview-1206": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
+        },
         "gemini-2.0-flash": {
             "prompt_short": 0.000_000_10,
             "completion_short": 0.000_000_40,
             "prompt_long": 0.000_000_10,
             "completion_long": 0.000_000_40,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-2.0-flash-latest": {
+            "prompt_short": 0.000_000_10,
+            "completion_short": 0.000_000_40,
+            "prompt_long": 0.000_000_10,
+            "completion_long": 0.000_000_40,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-2.0-flash-001": {
+            "prompt_short": 0.000_000_10,
+            "completion_short": 0.000_000_40,
+            "prompt_long": 0.000_000_10,
+            "completion_long": 0.000_000_40,
+            "cached": 0.000_000_037_5,
         },
         "gemini-2.0-flash-lite": {
             "prompt_short": 0.000_000_075,
             "completion_short": 0.000_000_30,
             "prompt_long": 0.000_000_075,
             "completion_long": 0.000_000_30,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-2.0-flash-lite-preview-02-05": {
+            "prompt_short": 0.000_000_075,
+            "completion_short": 0.000_000_30,
+            "prompt_long": 0.000_000_075,
+            "completion_long": 0.000_000_30,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-1.5-pro": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
+        },
+        "gemini-1.5-pro-latest": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
+        },
+        "gemini-1.5-pro-001": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
+        },
+        "gemini-1.5-pro-002": {
+            "prompt_short": 0.000_001_25,
+            "completion_short": 0.000_005,
+            "prompt_long": 0.000_002_5,
+            "completion_long": 0.000_01,
+            "cached": 0.000_000_625,
         },
         "gemini-1.5-flash": {
             "prompt_short": 0.000_000_075,
             "completion_short": 0.000_000_30,
             "prompt_long": 0.000_000_15,
             "completion_long": 0.000_000_60,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-1.5-flash-latest": {
+            "prompt_short": 0.000_000_075,
+            "completion_short": 0.000_000_30,
+            "prompt_long": 0.000_000_15,
+            "completion_long": 0.000_000_60,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-1.5-flash-001": {
+            "prompt_short": 0.000_000_075,
+            "completion_short": 0.000_000_30,
+            "prompt_long": 0.000_000_15,
+            "completion_long": 0.000_000_60,
+            "cached": 0.000_000_037_5,
+        },
+        "gemini-1.5-flash-002": {
+            "prompt_short": 0.000_000_075,
+            "completion_short": 0.000_000_30,
+            "prompt_long": 0.000_000_15,
+            "completion_long": 0.000_000_60,
+            "cached": 0.000_000_037_5,
         },
         "gemini-1.5-flash-8b": {
             "prompt_short": 0.000_000_037_5,
             "completion_short": 0.000_000_15,
             "prompt_long": 0.000_000_075,
             "completion_long": 0.000_000_30,
+            "cached": 0.000_000_025,
         },
-        "gemini-1.5-pro": {
-            "prompt_short": 0.000_001_25,
-            "completion_short": 0.000_005,
-            "prompt_long": 0.000_002_5,
-            "completion_long": 0.000_01,
+        "gemini-1.5-flash-8b-latest": {
+            "prompt_short": 0.000_000_037_5,
+            "completion_short": 0.000_000_15,
+            "prompt_long": 0.000_000_075,
+            "completion_long": 0.000_000_30,
+            "cached": 0.000_000_025,
+        },
+        "gemini-1.5-flash-8b-001": {
+            "prompt_short": 0.000_000_037_5,
+            "completion_short": 0.000_000_15,
+            "prompt_long": 0.000_000_075,
+            "completion_long": 0.000_000_30,
+            "cached": 0.000_000_025,
+        },
+        "gemini-1.5-flash-8b-002": {
+            "prompt_short": 0.000_000_037_5,
+            "completion_short": 0.000_000_15,
+            "prompt_long": 0.000_000_075,
+            "completion_long": 0.000_000_30,
+            "cached": 0.000_000_025,
         },
         "gemini-1.0-pro": {
             "prompt_short": 0.000_000_5,
             "completion_short": 0.000_001_5,
             "prompt_long": 0.000_000_5,
             "completion_long": 0.000_001_5,
+            "cached": 0.000_000,
         },
     }
 
     if input_tokens is None or output_tokens is None:
         return None
 
+    if cached_tokens is None:
+        cached_tokens = 0
+
     try:
         model_pricing = pricing[model]
     except KeyError:
@@ -77,12 +202,14 @@ def calculate_cost(
     use_long_context = input_tokens > 128_000
 
     prompt_price = model_pricing["prompt_long" if use_long_context else "prompt_short"]
+    cached_price = model_pricing["cached"]
     completion_price = model_pricing[
         "completion_long" if use_long_context else "completion_short"
     ]
 
     prompt_cost = input_tokens * prompt_price
+    cached_cost = cached_tokens * cached_price
     completion_cost = output_tokens * completion_price
-    total_cost = prompt_cost + completion_cost
+    total_cost = prompt_cost + cached_cost + completion_cost
 
     return total_cost

mirascope/core/google/_utils/_convert_message_params.py

@@ -16,50 +16,22 @@ from google.genai.types import (
 from ...base import BaseMessageParam
 from ...base._utils import get_audio_type, get_image_type
 from ...base._utils._parse_content_template import _load_media
-
-
-def _check_image_media_type(media_type: str) -> None:
-    """Raises a `ValueError` if the image media type is not supported."""
-    if media_type not in [
-        "image/jpeg",
-        "image/png",
-        "image/webp",
-        "image/heic",
-        "image/heif",
-    ]:
-        raise ValueError(
-            f"Unsupported image media type: {media_type}. "
-            "Google currently only supports JPEG, PNG, WebP, HEIC, "
-            "and HEIF images."
-        )
-
-
-def _check_audio_media_type(media_type: str) -> None:
-    """Raises a `ValueError` if the audio media type is not supported."""
-    if media_type not in [
-        "audio/wav",
-        "audio/mp3",
-        "audio/aiff",
-        "audio/aac",
-        "audio/ogg",
-        "audio/flac",
-    ]:
-        raise ValueError(
-            f"Unsupported audio media type: {media_type}. "
-            "Google currently only supports WAV, MP3, AIFF, AAC, OGG, "
-            "and FLAC audio file types."
-        )
+from ._validate_media_type import _check_audio_media_type, _check_image_media_type
 
 
 def _over_file_size_limit(size: int) -> bool:
-    """Check if the total file size exceeds the limit (15mb)."""
-    return size > 15 * 1024 * 1024  # 15MB
+    """Check if the total file size exceeds the limit (10mb).
+
+    Google limit is 20MB but base64 adds 33% to the size.
+    """
+    return size > 10 * 1024 * 1024  # 10MB
 
 
 async def _convert_message_params_async(
     message_params: list[BaseMessageParam | ContentDict], client: Client
 ) -> list[ContentDict]:
     converted_message_params = []
+    total_payload_size = 0
     for message_param in message_params:
         if not isinstance(message_param, BaseMessageParam):
             converted_message_params.append(message_param)
@@ -83,7 +55,6 @@ async def _convert_message_params_async(
             )
         else:
             converted_content = []
-            total_file_size = 0
             must_upload: dict[int, BlobDict] = {}
             for index, part in enumerate(content):
                 if part.type == "text":
@@ -93,10 +64,10 @@ async def _convert_message_params_async(
                     blob_dict = BlobDict(data=part.image, mime_type=part.media_type)
                     converted_content.append(PartDict(inline_data=blob_dict))
                     image_size = len(part.image)
-                    total_file_size += image_size
-                    if _over_file_size_limit(total_file_size):
+                    total_payload_size += image_size
+                    if _over_file_size_limit(total_payload_size):
                         must_upload[index] = blob_dict
-                        total_file_size -= image_size
+                        total_payload_size -= image_size
                 elif part.type == "image_url":
                     if (
                         client.vertexai
@@ -112,17 +83,17 @@ async def _convert_message_params_async(
                         )
                     else:
                         downloaded_image = _load_media(part.url)
-                        media_type = get_image_type(downloaded_image)
+                        media_type = f"image/{get_image_type(downloaded_image)}"
                         _check_image_media_type(media_type)
                         blob_dict = BlobDict(
                             data=downloaded_image, mime_type=media_type
                         )
                         converted_content.append(PartDict(inline_data=blob_dict))
                         image_size = len(downloaded_image)
-                        total_file_size += image_size
-                        if _over_file_size_limit(total_file_size):
+                        total_payload_size += image_size
+                        if _over_file_size_limit(total_payload_size):
                             must_upload[index] = blob_dict
-                            total_file_size -= image_size
+                            total_payload_size -= image_size
                 elif part.type == "audio":
                     _check_audio_media_type(part.media_type)
                     audio_data = (
@@ -133,10 +104,10 @@ async def _convert_message_params_async(
                     blob_dict = BlobDict(data=audio_data, mime_type=part.media_type)
                     converted_content.append(PartDict(inline_data=blob_dict))
                     audio_size = len(audio_data)
-                    total_file_size += audio_size
-                    if _over_file_size_limit(total_file_size):
+                    total_payload_size += audio_size
+                    if _over_file_size_limit(total_payload_size):
                         must_upload[index] = blob_dict
-                        total_file_size -= audio_size
+                        total_payload_size -= audio_size
                 elif part.type == "audio_url":
                     if (
                         client.vertexai
@@ -152,17 +123,17 @@ async def _convert_message_params_async(
                         )
                     else:
                         downloaded_audio = _load_media(part.url)
-                        media_type = get_audio_type(downloaded_audio)
+                        media_type = f"audio/{get_audio_type(downloaded_audio)}"
                         _check_audio_media_type(media_type)
                         blob_dict = BlobDict(
                             data=downloaded_audio, mime_type=media_type
                         )
                         converted_content.append(PartDict(inline_data=blob_dict))
                         audio_size = len(downloaded_audio)
-                        total_file_size += audio_size
-                        if _over_file_size_limit(total_file_size):
+                        total_payload_size += audio_size
+                        if _over_file_size_limit(total_payload_size):
                             must_upload[index] = blob_dict
-                            total_file_size -= audio_size
+                            total_payload_size -= audio_size
                 else:
                     raise ValueError(
                         "Google currently only supports text, image, and audio parts. "
@@ -211,4 +182,5 @@ def convert_message_params(
             )
             return future.result()
     except RuntimeError:
-        return asyncio.run(_convert_message_params_async(message_params, client))
+        ...
+    return asyncio.run(_convert_message_params_async(message_params, client))

mirascope/core/google/_utils/_message_param_converter.py

@@ -13,35 +13,15 @@ from mirascope.core.base._utils._base_message_param_converter import (
     BaseMessageParamConverter,
 )
 from mirascope.core.base.message_param import (
+    AudioPart,
     AudioURLPart,
     ImageURLPart,
     ToolCallPart,
     ToolResultPart,
 )
-from mirascope.core.gemini._utils._message_param_converter import _is_audio_mime
 from mirascope.core.google._utils import convert_message_params
 
-
-def _is_image_mime(mime_type: str) -> bool:
-    return mime_type in ["image/jpeg", "image/png", "image/gif", "image/webp"]
-
-
-def _to_image_part(mime_type: str, data: bytes) -> ImagePart:
-    if not _is_image_mime(mime_type):
-        raise ValueError(
-            f"Unsupported image media type: {mime_type}. "
-            "Expected one of: image/jpeg, image/png, image/gif, image/webp."
-        )
-    return ImagePart(type="image", media_type=mime_type, image=data, detail=None)
-
-
-def _to_document_part(mime_type: str, data: bytes) -> DocumentPart:
-    if mime_type != "application/pdf":
-        raise ValueError(
-            f"Unsupported document media type: {mime_type}. "
-            "Only application/pdf is supported."
-        )
-    return DocumentPart(type="document", media_type=mime_type, document=data)
+from ._validate_media_type import _check_audio_media_type, _check_image_media_type
 
 
 class GoogleMessageParamConverter(BaseMessageParamConverter):
@@ -74,21 +54,42 @@ class GoogleMessageParamConverter(BaseMessageParamConverter):
                 if part.text:
                     content_list.append(TextPart(type="text", text=part.text))
 
-                elif part.inline_data:
-                    blob = part.inline_data
-                    mime = blob.mime_type or ""
+                elif blob := part.inline_data:
+                    mime_type = blob.mime_type or ""
                     data = blob.data or b""
-                    if _is_image_mime(mime or ""):
-                        content_list.append(_to_image_part(mime, data))
-                    elif mime == "application/pdf":
-                        content_list.append(_to_document_part(mime, data))
+                    if mime_type.startswith("image/"):
+                        _check_image_media_type(mime_type)
+                        content_list.append(
+                            ImagePart(
+                                type="image",
+                                media_type=mime_type,
+                                image=data,
+                                detail=None,
+                            )
+                        )
+                    elif mime_type.startswith("audio/"):
+                        _check_audio_media_type(mime_type)
+                        content_list.append(
+                            AudioPart(
+                                type="audio",
+                                media_type=mime_type,
+                                audio=data,
+                            )
+                        )
+                    elif mime_type == "application/pdf":
+                        content_list.append(
+                            DocumentPart(
+                                type="document", media_type=mime_type, document=data
+                            )
+                        )
                     else:
                         raise ValueError(
-                            f"Unsupported inline_data mime type: {mime}. Cannot convert to BaseMessageParam."
+                            f"Unsupported inline_data mime type: {mime_type}. Cannot convert to BaseMessageParam."
                         )
 
-                elif part.file_data:
-                    if _is_image_mime(cast(str, part.file_data.mime_type)):
+                elif file_data := part.file_data:
+                    mime_type = file_data.mime_type or ""
+                    if mime_type.startswith("image/"):
                         content_list.append(
                             ImageURLPart(
                                 type="image_url",
@@ -96,7 +97,7 @@ class GoogleMessageParamConverter(BaseMessageParamConverter):
                                 detail=None,
                             )
                         )
-                    elif _is_audio_mime(cast(str, part.file_data.mime_type)):
+                    elif mime_type.startswith("audio/"):
                         content_list.append(
                             AudioURLPart(
                                 type="audio_url",

mirascope/core/google/_utils/_validate_media_type.py

@@ -0,0 +1,34 @@
+"""Utilities for validating supported media types for Google models."""
+
+
+def _check_image_media_type(media_type: str) -> None:
+    """Raises a `ValueError` if the image media type is not supported."""
+    if media_type not in [
+        "image/jpeg",
+        "image/png",
+        "image/webp",
+        "image/heic",
+        "image/heif",
+    ]:
+        raise ValueError(
+            f"Unsupported image media type: {media_type}. "
+            "Google currently only supports JPEG, PNG, WebP, HEIC, "
+            "and HEIF images."
+        )
+
+
+def _check_audio_media_type(media_type: str) -> None:
+    """Raises a `ValueError` if the audio media type is not supported."""
+    if media_type not in [
+        "audio/wav",
+        "audio/mp3",
+        "audio/aiff",
+        "audio/aac",
+        "audio/ogg",
+        "audio/flac",
+    ]:
+        raise ValueError(
+            f"Unsupported audio media type: {media_type}. "
+            "Google currently only supports WAV, MP3, AIFF, AAC, OGG, "
+            "and FLAC audio file types."
+        )

mirascope/core/google/call_response.py

@@ -11,6 +11,7 @@ from google.genai.types import (
     ContentListUnionDict,
     FunctionResponseDict,
     GenerateContentResponse,
+    GenerateContentResponseUsageMetadata,
     PartDict,
     # Import manually SchemaDict to avoid Pydantic error
     SchemaDict,  # noqa: F401
@@ -68,11 +69,13 @@ class GoogleCallResponse(
 
     _provider = "google"
 
+    @computed_field
     @property
     def content(self) -> str:
         """Returns the contained string content for the 0th choice."""
         return self.response.candidates[0].content.parts[0].text  # pyright: ignore [reportOptionalSubscript, reportReturnType, reportOptionalMemberAccess, reportOptionalIterable]
 
+    @computed_field
     @property
     def finish_reasons(self) -> list[str]:
         """Returns the finish reasons of the response."""
@@ -83,6 +86,7 @@ class GoogleCallResponse(
             if candidate and candidate.finish_reason is not None
         ]
 
+    @computed_field
     @property
     def model(self) -> str:
         """Returns the model name.
@@ -90,8 +94,11 @@ class GoogleCallResponse(
         google.generativeai does not return model, so we return the model provided by
         the user.
         """
-        return self._model
+        return (
+            self.response.model_version if self.response.model_version else self._model
+        )
 
+    @computed_field
     @property
     def id(self) -> str | None:
         """Returns the id of the response.
@@ -101,13 +108,14 @@ class GoogleCallResponse(
         return None
 
     @property
-    def usage(self) -> None:
+    def usage(self) -> GenerateContentResponseUsageMetadata | None:
         """Returns the usage of the chat completion.
 
         google.generativeai does not have Usage, so we return None
         """
-        return None
+        return self.response.usage_metadata
 
+    @computed_field
     @property
     def input_tokens(self) -> int | None:
         """Returns the number of input tokens."""
@@ -117,6 +125,17 @@ class GoogleCallResponse(
             else None
         )
 
+    @computed_field
+    @property
+    def cached_tokens(self) -> int | None:
+        """Returns the number of cached tokens."""
+        return (
+            self.response.usage_metadata.cached_content_token_count
+            if self.response.usage_metadata
+            else None
+        )
+
+    @computed_field
     @property
     def output_tokens(self) -> int | None:
         """Returns the number of output tokens."""
@@ -126,10 +145,13 @@ class GoogleCallResponse(
             else None
         )
 
+    @computed_field
     @property
     def cost(self) -> float | None:
         """Returns the cost of the call."""
-        return calculate_cost(self.input_tokens, self.output_tokens, self.model)
+        return calculate_cost(
+            self.input_tokens, self.cached_tokens, self.output_tokens, self.model
+        )
 
     @computed_field
     @cached_property