python-genai

1.1.0 (2025-02-10)

Features

• Add support for typing.Literal in response_schema (fixes #264) (384c4eb)
• Allow converting a function into FunctionDeclaration with api option (408e28f)
• Support generate content config for each chat turn (ca19100)
• Tuning - Tuning.tune for Gemini API no longer blocks until the tuning operation is resolved (fcf8888)

Bug Fixes

• Remove duplicate function invocations and ensure automatic function calling can be disabled in generate_content_stream (0958fbe)
• Response_schema generation for pydantic fields with type Optional[list] and Optional[MyBaseModel] now work correctly (fixes #246) (8330561)
• Support dict response_schema with 'any_of' key (75f5056)
• Common - Do not fail when server returns unknown enum values on Python <3.11 (843d86d)

Documentation

• Add docs on how to structure contents (fixes #274) (da356cb)
• Regenerate docs for 1.0 (fbee816)
• Update model names in README, fix function calling example (11c0274)

1.0.0 (2025-02-05)

⚠ BREAKING CHANGES

• Remove deprecated field: deprecated_response_payload

Features

• Add labels for GenerateContent requests (3e3b82d)
• Add Union support to response_schema (eedf4f1)
• Support automatic function calling in models.generate_content_stream and send_message_stream, sync and async mode (7c9f8b5)

Bug Fixes

• Avoid false test failure by skipping incompatible test case for python 3.13 (7f10e55)
• Default to list base models (instead of tuned models) (ef48f0d)
• Handle pydantic model type recognition gracefully for python 3.9 and 3.10 (4a38fdf)
• Raise error when Gemini API response_schema has 'default' or 'anyof' fields (c50bca0)
• Remove redundant contents in Automatic Function Calling history (5510595)
• Remove unsupported parameter from Gemini API (f8addb5)

Documentation

• Remove experimental classification from description. (b14ac57)
• Remove thoughts examples and update tests (af3b339)
• Update documentation on how to set api version using genai client (6fd4425)
• Update instruction on function calling experience in ANY mode. (451cf98)

Code Refactoring

• Remove deprecated field: deprecated_response_payload (197fa46)

Features

• Add labels for GenerateContent requests (3e3b82d)
• Add Union support to response_schema (eedf4f1)

Bug Fixes

• Avoid false test failure by skipping incompatible test case for python 3.13 (7f10e55)
• Default to list base models (instead of tuned models) (ef48f0d)
• Raise error when Gemini API response_schema has 'default' or 'anyof' fields (c50bca0)
• Remove redundant contents in Automatic Function Calling history (5510595)

Documentation

• Remove thoughts examples and update tests (af3b339)

0.8.0 (2025-01-30)

⚠ BREAKING CHANGES

• Rename files.upload argument to "file" instead of "path".

Features

• Add enhanced_prompt to GeneratedImage class (76c810b)
• Added Operation and PredictOperation (internal module) (309dd26)
• Support global endpoint natively in Vertex (f4530b0)
• Support unknown enum values (da448b3)

Bug Fixes

• Streaming in Vertex AI Express (ff78b7b)

Documentation

• Correct generate content with uploaded file example (8cea052)

Miscellaneous Chores

• Rename files.upload argument to "file" instead of "path". (f68aa1f)

0.7.0 (2025-01-28)

⚠ BREAKING CHANGES

• Remove skip_project_and_location_in_path from async models.list
• remove "tuning.distill"
• Change asynchronous streaming output to an awaitable async iterator for client.aio.models.generate_content_stream and client.aio.chat.send_message_stream
• Remove pillow as a required dependency
• rename batches.list method signature.
• make Part, FunctionDeclaration, Image, and GenerateContentResponse classmethods argument keyword only
• Remove skip_project_and_location_in_path from HttpOption
• Renamed FunctionDeclaration class functions to reflect the fact that they work off of the Callable type not just functions.
• Moved the HttpOptions class into types.py, by making it autogenerated. This causes the following breaking change:
• rename generate_image() to generate_images(), rename GenerateImageConfig to GenerateImagesConfig, rename GenerateImageResponse to GenerateImagesResponse, rename GenerateImageParameters to GenerateImagesParameters

Features

• [genai-modules][models] Add HttpOptions to all method configs for models. (76fdde7)
• Add support for enhance_prompt to model.generate_image (d09e14e)
• Added support for the new HttpOptions class in the per request options overrides. (ad57025)
• Change asynchronous streaming output to an awaitable async iterator for client.aio.models.generate_content_stream and client.aio.chat.send_message_stream (0c124eb)
• Enable enum support in the GenerateContentConfig.response_schema (fe82e10)
• Handle a wider variety of response_schemas - primitives and nested lists. (24fffea)
• Images - Added Image.mime_type (71a8f1d)
• Make Part, FunctionDeclaration, Image, and GenerateContentResponse classmethods argument keyword only (b58f4e0)
• Remove skip_project_and_location_in_path from async models.list (d788626)
• Remove skip_project_and_location_in_path from HttpOption (a1c0435)
• Support GenerateContentResponse.parsed to return Enum (e214211)
• Validate enum value for different backend endpoint. (97bb958)

Bug Fixes

• Add required key to nested fields in function declaration schemas (c7dba47)
• Fix pydantic list support in Python 3.9 and 3.10 (fixes #52) (81150b3)
• Handle nullable types in pydantic classes (fixes #62) (773e1c0)
• Support parameterized generics Union type in automatic function calling parameters (fixes #22) (04475ab)

Documentation

• Add examples & tests for thinking model (59e2763)
• Correct usage (0c546de)
• Document experimental state of the live module. (115ac47)
• Fix Blob type docstring. (7f38e55)
• Remove repeated docstring for interrupted in docs (230cfbc)
• Remove the word "class" in docstring and sync Live types docstring up-to-date (7fa3ef3)
• Update supported model parameter format for generate_content (a5c3a1c)

Code Refactoring

• Moved the HttpOptions class into types.py, by making it autogenerated. This causes the following breaking change: (ad57025)
• Remove "tuning.distill" (ca8cd5e)
• Remove pillow as a required dependency (8ccdf2e)
• Rename batches.list method signature. (aa7c071)
• Rename generate_image() to generate_images(), rename GenerateImageConfig to GenerateImagesConfig, rename GenerateImageResponse to GenerateImagesResponse, rename GenerateImageParameters to GenerateImagesParameters (65d7bf5)
• Renamed FunctionDeclaration class functions to reflect the fact that they work off of the Callable type not just functions. (0e4a003)

0.6.0 (2025-01-21)

⚠ BREAKING CHANGES

• Renamed types.FunctionDeclaration.from_function to types.FunctionDeclaration.from_callable (5afa6bb)

Features

• Add support for audio_timestamp to types.GenerateContentConfig (fixes #132) (116395b)
• Add support for case insensitive enum types. (fixes #11) (252f302)
• Add support for class methods in the Tools for the GenerateContent Function (5afa6bb)
• Add ThinkingConfig to generate content config. (c23c42c)
• Implement client.files.download (a034b77)
• Support BytesIO when uploading files (4b490dc)
• Support lists in response_schema (8ed933d)
• Usability change to simplify generate content with uploaded files (5c372ae)

Bug Fixes

• Fix count_tokens system_instruction and tools config (056eaba)
• Fix models.list() with empty tuned models (75409f1)
• Fixed the bytes type handling (the base64.urlsafe_bencode error) (1bc161d)
• Project and location are required when using client in Vertex AI mode. (d5859fa)

Documentation

• Add project description to README. (384f413)
• Update formatting in README.md (d33fb37)
• Update models.list doc and code examples to include list base models (f2e0c43)
• Tool use example in docs (87fe5b0)
• Tool use example in README.md (87fe5b0)

0.5.0 (2025-01-13)

Features

• Support API keys for VertexAI mode generate_content (0e4b0e5)
• Support list models to return base models (0f713f1)
• Support parsing 'interrupted' field in Live Python SDK. (eab2c4a)
• Use ser_json_byte val_json_bytes in bytes type public interface (1176e43)

Bug Fixes

• Update header type (62c45f9)

Documentation

• Correct description of path parameter that only path-like object is supported (336498b)

0.4.0 (2025-01-08)

⚠ BREAKING CHANGES

• Make Imagen upscale_factor a required argument, make upscale config optional

Features

• ApiClient - support timeout in HttpOptions (b22286f)
• Enable response_logprobs and logprobs for Google AI (5586f3d)
• Support function_calls quick accessor in GenerateContentResponse class (81b8a23)

Bug Fixes

• Change string type to int type (5c15243)
• FunctionCallCancellation ids type. (b0e46b7)
• Gracefully catch error if streamed json does not meet schema validation (fixes #14) (f494432)
• Update RealtimeClientLiveMessage realtime content parameter field. (4340939)

Documentation

• Add readme example for Chats module (830f0f5)
• Fix typo in code document (f9243e8)
• Fix typo in CONTRIBUTING.md (3e42644)

Code Refactoring

• Make Imagen upscale_factor a required argument, make upscale config optional (b06629f)

0.3.0 (2024-12-17)

BREAKING CHANGES

• contents must be passed to CreateCachedContentConfig instead of as a parameter to create_cached_content.

Features

• Add support for Pydantic default value in Automatic Function Calling.
• Add support for thought.
• Add support for streaming chat.