Add support for Gemini 3 Pro Preview

Author: DouweM

docs/models/google.md

@@ -214,22 +214,22 @@ from pydantic_ai.models.google import GoogleModel, GoogleModelSettings
 settings = GoogleModelSettings(
     temperature=0.2,
     max_tokens=1024,
-    google_thinking_config={'thinking_budget': 2048},
+    google_thinking_config={'thinking_level': 'low'},
     google_safety_settings=[
         {
             'category': HarmCategory.HARM_CATEGORY_HATE_SPEECH,
             'threshold': HarmBlockThreshold.BLOCK_LOW_AND_ABOVE,
         }
     ]
 )
-model = GoogleModel('gemini-2.5-flash')
+model = GoogleModel('gemini-2.5-pro')
 agent = Agent(model, model_settings=settings)
 ...
 ```
 
 ### Disable thinking
 
-You can disable thinking by setting the `thinking_budget` to `0` on the `google_thinking_config`:
+On models older than Gemini 2.5 Pro, you can disable thinking by setting the `thinking_budget` to `0` on the `google_thinking_config`:
 
 ```python
 from pydantic_ai import Agent

pydantic_ai_slim/pydantic_ai/__init__.py

@@ -46,6 +46,7 @@
     DocumentFormat,
     DocumentMediaType,
     DocumentUrl,
+    FileOptions,
     FilePart,
     FileUrl,
     FinalResultEvent,
@@ -55,6 +56,7 @@
     HandleResponseEvent,
     ImageFormat,
     ImageMediaType,
+    ImageOptions,
     ImageUrl,
     ModelMessage,
     ModelMessagesTypeAdapter,
@@ -82,6 +84,7 @@
     UserPromptPart,
     VideoFormat,
     VideoMediaType,
+    VideoOptions,
     VideoUrl,
 )
 from .output import NativeOutput, PromptedOutput, StructuredDict, TextOutput, ToolOutput
@@ -146,6 +149,7 @@
     'DocumentFormat',
     'DocumentMediaType',
     'DocumentUrl',
+    'FileOptions',
     'FileUrl',
     'FilePart',
     'FinalResultEvent',
@@ -155,6 +159,7 @@
     'HandleResponseEvent',
     'ImageFormat',
     'ImageMediaType',
+    'ImageOptions',
     'ImageUrl',
     'BinaryImage',
     'ModelMessage',
@@ -183,6 +188,7 @@
     'UserPromptPart',
     'VideoFormat',
     'VideoMediaType',
+    'VideoOptions',
     'VideoUrl',
     # profiles
     'ModelProfile',

pydantic_ai_slim/pydantic_ai/messages.py

@@ -13,7 +13,7 @@
 import pydantic_core
 from genai_prices import calc_price, types as genai_types
 from opentelemetry._events import Event  # pyright: ignore[reportPrivateImportUsage]
-from typing_extensions import deprecated
+from typing_extensions import TypedDict, deprecated
 
 from . import _otel_messages, _utils
 from ._utils import generate_tool_call_id as _generate_tool_call_id, now_utc as _now_utc
@@ -106,6 +106,61 @@ def _multi_modal_content_identifier(identifier: str | bytes) -> str:
     return hashlib.sha1(identifier).hexdigest()[:6]
 
 
+class FileOptions(TypedDict, total=False):
+    """Options for how the provider should process the file."""
+
+    pass
+
+
+class VideoOptions(TypedDict, total=False):
+    """Options for how the provider should process the video."""
+
+    detail: Literal['high', 'medium', 'low']
+    """The detail level of the video.
+
+    Supported by:
+
+    - Google: Maps to `media_resolution`. # TODO (DouweM): URL
+    """
+
+    fps: float
+    """The frame rate of the video sent to the model. If not specified, the default value will be 1.0. The fps range is (0.0, 24.0].
+
+    Supported by:
+
+    - Google: https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing
+    """
+
+    start_offset: str
+    """The start offset of the video sent to the model.
+
+    Supported by:
+
+    - Google: https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing
+    """
+
+    end_offset: str
+    """The end offset of the video sent to the model.
+
+    Supported by:
+
+    - Google: https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing
+    """
+
+
+class ImageOptions(TypedDict, total=False):
+    """Options for how the provider should process the image."""
+
+    detail: Literal['high', 'medium', 'low']
+    """The detail level of the image.
+
+    Supported by:
+
+    - OpenAI: Supports only `high` and `low`. https://platform.openai.com/docs/guides/images-vision?api-mode=responses#specify-image-input-detail-level
+    - Google: Maps to `media_resolution`. # TODO (DouweM): URL
+    """
+
+
 @dataclass(init=False, repr=False)
 class FileUrl(ABC):
     """Abstract base class for any URL-based file."""
@@ -122,13 +177,9 @@ class FileUrl(ABC):
     * If False, the URL is sent directly to the model and no download is performed.
     """
 
-    vendor_metadata: dict[str, Any] | None = None
-    """Vendor-specific metadata for the file.
-
-    Supported by:
-    - `GoogleModel`: `VideoUrl.vendor_metadata` is used as `video_metadata`: https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing
-    - `OpenAIChatModel`, `OpenAIResponsesModel`: `ImageUrl.vendor_metadata['detail']` is used as `detail` setting for images
-    """
+    # TODO (v2): Rename to `options`?
+    vendor_metadata: FileOptions | None = None
+    """Options on how the provider should process the file."""
 
     _media_type: Annotated[str | None, pydantic.Field(alias='media_type', default=None, exclude=True)] = field(
         compare=False, default=None
@@ -145,7 +196,7 @@ def __init__(
         media_type: str | None = None,
         identifier: str | None = None,
         force_download: bool = False,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: FileOptions | None = None,
     ) -> None:
         self.url = url
         self._media_type = media_type
@@ -209,7 +260,7 @@ def __init__(
         media_type: str | None = None,
         identifier: str | None = None,
         force_download: bool = False,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: VideoOptions | None = None,
         kind: Literal['video-url'] = 'video-url',
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
@@ -285,7 +336,7 @@ def __init__(
         media_type: str | None = None,
         identifier: str | None = None,
         force_download: bool = False,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: FileOptions | None = None,
         kind: Literal['audio-url'] = 'audio-url',
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
@@ -348,7 +399,7 @@ def __init__(
         media_type: str | None = None,
         identifier: str | None = None,
         force_download: bool = False,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: ImageOptions | None = None,
         kind: Literal['image-url'] = 'image-url',
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
@@ -406,7 +457,7 @@ def __init__(
         media_type: str | None = None,
         identifier: str | None = None,
         force_download: bool = False,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: FileOptions | None = None,
         kind: Literal['document-url'] = 'document-url',
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _media_type: str | None = None,
@@ -476,12 +527,8 @@ class BinaryContent:
     media_type: AudioMediaType | ImageMediaType | DocumentMediaType | str
     """The media type of the binary data."""
 
-    vendor_metadata: dict[str, Any] | None = None
-    """Vendor-specific metadata for the file.
-
-    Supported by:
-    - `GoogleModel`: `BinaryContent.vendor_metadata` is used as `video_metadata`: https://ai.google.dev/gemini-api/docs/video-understanding#customize-video-processing
-    - `OpenAIChatModel`, `OpenAIResponsesModel`: `BinaryContent.vendor_metadata['detail']` is used as `detail` setting for images
+    vendor_metadata: FileOptions | None = None
+    """Options on how the provider should process the file.
     """
 
     _identifier: Annotated[str | None, pydantic.Field(alias='identifier', default=None, exclude=True)] = field(
@@ -491,13 +538,52 @@ class BinaryContent:
     kind: Literal['binary'] = 'binary'
     """Type identifier, this is available on all parts as a discriminator."""
 
+    @overload
+    def __init__(
+        self,
+        data: bytes,
+        *,
+        media_type: ImageMediaType,
+        identifier: str | None = None,
+        vendor_metadata: ImageOptions | None = None,
+        kind: Literal['binary'] = 'binary',
+        # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
+        _identifier: str | None = None,
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        data: bytes,
+        *,
+        media_type: VideoMediaType,
+        identifier: str | None = None,
+        vendor_metadata: VideoOptions | None = None,
+        kind: Literal['binary'] = 'binary',
+        # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
+        _identifier: str | None = None,
+    ) -> None: ...
+
+    @overload
+    def __init__(
+        self,
+        data: bytes,
+        *,
+        media_type: AudioMediaType | DocumentMediaType | str,
+        identifier: str | None = None,
+        vendor_metadata: FileOptions | None = None,
+        kind: Literal['binary'] = 'binary',
+        # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
+        _identifier: str | None = None,
+    ) -> None: ...
+
     def __init__(
         self,
         data: bytes,
         *,
-        media_type: AudioMediaType | ImageMediaType | DocumentMediaType | str,
+        media_type: AudioMediaType | ImageMediaType | VideoMediaType | DocumentMediaType | str,
         identifier: str | None = None,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: FileOptions | None = None,
         kind: Literal['binary'] = 'binary',
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         _identifier: str | None = None,
@@ -516,7 +602,7 @@ def narrow_type(bc: BinaryContent) -> BinaryContent | BinaryImage:
                 data=bc.data,
                 media_type=bc.media_type,
                 identifier=bc.identifier,
-                vendor_metadata=bc.vendor_metadata,
+                vendor_metadata=cast(ImageOptions, bc.vendor_metadata),
             )
         else:
             return bc
@@ -599,7 +685,7 @@ def __init__(
         *,
         media_type: str,
         identifier: str | None = None,
-        vendor_metadata: dict[str, Any] | None = None,
+        vendor_metadata: ImageOptions | None = None,
         # Required for inline-snapshot which expects all dataclass `__init__` methods to take all field names as kwargs.
         kind: Literal['binary'] = 'binary',
         _identifier: str | None = None,