libs/vertexai/v3.0.0rc1

Initial release

increment versions
update deps
lock
feat(vertex): support standard content (#1248)
Merge branch 'main' into v3.0.0
fix(vertex): skip tests for retired models (#1247)
Merge branch 'main' into v3.0.0
fix(vertex): ToolMessage is translated to function role which does not exist . (#1238)
pass on cleaning up vertex (#1244)
Merge branch 'main' into v3.0.0
Revert "feat(vertex): standard content" (#1240)
feat(vertex): standard content (#1224)
feat(vertex): migrated embeddings to the new SDK (#1229)
test: use gemini-flash-lite-latest (#1230)
vertex lock update, no version bump
Merge branch 'main' into v3.0.0
chore(genai): update from gemini-1.5-flash to gemini-2.5-flash (#1221)
chore(vertex): remove Gemma integrations (#1220)
!chore(vertex): remove deprecated 'embbedings' parameter and warning
!chore(vertex): remove embbedings property
!chore(vertex): remove embed_image()
resolve: lint
chore: update projects/locks
Merge branch 'main' into v3.0.0
chore: bump lockfiles, resolve linting (genai, vertex, community) (#1203)
Merge branch 'main' into v3.0.0
feat(vertex): add additional args for vertex ai vector search (#1198)
fix: resolve VertexCandidate type annotation
Merge branch 'main' into v3.0.0
chore(vertex): add migration TODOs, suppress warning (#1196)
fix(vertex): lockfile
chore: v1 migrations (#1195)
chore: update min vers in pyproject.toml files
chore: linting and updating typing as a result of upgrade
chore: lower bound to Python 3.10, keep formatting py39
fix(vertex): honor max_retries=0 and disable GAPIC retries (#1185)
fix(vertex): update model version check for text embedding (#1184)
genai, vertex: nits (#1173)
fix: use .invoke() instead of __call__ for LLMs (#1172)
release(vertexai): 2.1.2 (#1161)
refactor(vertex): resolve test warnings (#1166)
fix(vertexai): handle unrecognized FinishReason values (#1164)
fix(vertexai): update audio timestamp test regex pattern (#1163)
chore(vertex): update embedding models to use 'gemini-embedding-001' and 'imagen-3.0' (#1086)
fix(vertex): use imagen-3.0-generate-002 as default model for image generation (#1162)
VectorSearchVectorStore: integer handling in numeric restrictions (#1155)
Guard against unknown FinishReason values (#1085)
fix(vertexai): add Union type support to V1 schema formatter (#1137)
vertex: add http_client support to ChatAnthropicVertex (#1154)
release(vertexai): 2.1.1 (#1149)
vertex: fix AIMessage tool call handling with empty content (#1145)
vertexai(patch): fix function calling for v1 endpoints (#1147)
vertexai: handle messages that evaluate to a literal float (#938)
chore(tests): add vertex type ignore specifiers (#1134)
fix(vertexai): Set requires_reference=True for VertexPairWiseStringEvaluator (#1133)
vertex: release 2.1.0 (#1132)
feat: add VSCode settings for Ruff fixes (#1128)
test(vertexai): add pytest warnings filter for unsupported parameters in ChatVertexAI (#1129)
test(vertexai): suppress UserWarning in parameter tests (#1130)
vertexai: require opt-in for ast.literal_eval on input message strings (#1122)
fix(vertexai): resolve Pydantic deprecation warning in VertexAI LLM (#1127)
vertexai: resolve errors from uv migration (#1125)
infra: poetry -> uv (#1101)
vertexai: Added ChatVertexAI test case with configurable_fields for thinking_budget (#926)
fix: Add cache token support to ChatAnthropicVertex streaming responses (#1010)
vertexai/genai: Added missing chunk to on_llm_new_token to _stream_gemini (#1058)
[vertexai] docs: enhance multimodal input examples for ChatVertexAI (#1074)
release 2.0.28 (#1073)
vertexai: fix(ChatAnthropicVertex): support BlockType Image (#1070)
vertexai : add media_resolution parameter support to ChatVertexAI (#1066)
vertexai: fix handling of single string in add_texts (#860)
vertexai: Added Image Generation Support (#1054)
vertexai: Add support for image output in ChatVertexAI code execution (#1051)
vertexai: fixes floats parsed and assumed iterable (#1018)
[genai, vertexai]: Add anyOf support for Union types in tool calling (#992)
fix: update thinking model name to match latest version (#1044)
chore: added latest meta models (#1036)
[vertexai] Fix pydantic deprecation warning: Use class-level model_fields (#996)
cr
support for function calls
add tests
docs: fixes langchain-ai/langchain#31785 (#1017)
vertexai: release 2.0.27
vertexai: add support for include_thoughts parameter (#968)
genai: Add json_mode support for structured output and schema reference handling in ChatGoogleGenerativeAI (#970)
removed deprecated PALM models (#1001)
release 2.0.26 (#993)
feat: Add cached_content_token_count to lc_usage for vertexai (#990)
fix tests (#991)
vertexai: add support for gemini-embedding-001 (#955)
[vertexai]: Fix #845 tool binding issue with anyOf Field in pydantic v2 (#987)
feat: Pass base_url to AnthropicVertex (#983)
[vertexai]: ignore empty AIMessage from previous blocked responses (#929)
bump core and increment version (#979)
vertexai[patch]: support code_execution (#959)
[vertexai]: fix IndexError when a tool returns an empty list (#976)
vertexai: strip nullable anyOf when using dedicated structured output feature (#977)
vertexai[patch]: cache prediction client (#969)
vertexai[patch]: bump langchain-tests (#966)
vertexai: docs added comprehensive README for langchain-google-vertexai integration package (#934)
added total and thinking tokens to callback (#923)
release 2.0.24 (#924)
fixed thinking mode (#922)
Added audio_timestamp support. (#913)
release 2.0.23 (#909)
vertexai: Added thinking_budget support to ChatVertexAI (#893)
aiplatform 1.92.0 (#906)
Release 2.0.22 (#905)
bumped aiplatform version (#898)
vertexai: add frequency_penalty and presence_penalty (#889)
[genai|vertexai|community]: Adding Vertex Tool Context into client info (#875)
genai, vertexai[patch]: support model_kwargs (#891)
vertexai: Fix global region endpoint for ChatVertexAI (#867)
vertexai: release 2.0.21 (#888)
vertexai: loosen dependency on validators package (#871)
remove stream from params if passed (#880)
vertexai[patch]: add standard test for audio inputs (#870)
fix tests (#883)
upgraded aiplatform version (#876)
vertex[patch]: add support for google_search tool passed as a dict (#863)
vertexai[patch]: support standard format for multi-modal content blocks (#861)
doc(vertexai): typo in doc (#854)
vertexai: fix null ids in add_texts_with_embeddings (#856)
make test python 3.9 compatible (#855)
release 2.0.20 (#851)
Fix resolving of references in ChatVertexAI.with_structured_output (#843)
Fix bind tools (#849)
vertexai: fix typos in vectorstores (#844)
Fix (#841)
vertexai: Fixed content coercion while streaming Claude thinking response (#834)
fixed Anthropic url support (#839)
release 2.0.19 (#840)
vertexai: updating aiplatform (#827)
Docs: Add context caching examples (#835)
[genai, vertexai]: Add warnings for invalid kwargs (#829)
update (#830)
Fix: #802 - Allow passing google.oauth2 Credentials to vectorstores (#824)
bump deps (#818)
Fix incorrect import statement in ChatVertexAI documentation (#815)
vertexai[patch]: return model name in response metadata when streaming (#813)
bump core and increment version (#812)
vertex[patch]: support mistral-small-2503 in model garden (#793)
Add GCS image support in ChatAnthropicVertex model (#808)
vertexai: add support for llama-3.3-70b model in MAAS integration (#785)
remove warning (#810)
Add Image URL support in ChatAnthropicVertex model (#804)
Add model_kwargs to ChatAnthropicVertex to enable thinking mode (#797)
increment version (#801)
vertexai[patch]: support TypedDict and Pydantic v1 with json_mode (#799)
Add Chain of Thoughts support for ChatAnthropicVertex Claude 3.7 (#790)
vertexai: add custom wait time for retries (#778)
fix:enhance _tool_choice_to_tool_config to handle additional function format (#779)
Avoid retries for InvalidArgument exceptions (#777)
bugfix: use base_url if set (#774)
vertex[patch]: update langchain-tests (#773)
release 2.0.15 (#771)
Add support for "model_name" in the ChatResult (#770)
Bug Fix: Make call of literal_eval optional (#766)
fix: typo in error message (#768)
added endpoint_version (#767)
Adding Labels in langchain for gemini billing and tracebility (#732)
Enable passing schema to ChatVertexAI client (#763)
fixed default transport for async client (#760)
bumped sdk version and moved tests to extended (#757)
release 2.0.14 (#753)
fix with_structured_output with json_mode (#752)
add timeout (#750)
use ChatModel as a client for VertexAI (#749)
Ensure single instance of ImageBytesLoader in Embeddings (#737)
Ensure single instance of ImageBytesLoader in Vision Models (#736)
vertexai: add retry tracing with metadata in ChatAnthropicVertex (#728)
Ensure a single instance of ImageBytesLoader throughout chat_models (#722)
release 2.0.13 (#723)
vertexai: Add embeddings_task_type parameter to embed_query and embed_documents (#716)
Fix: #720 - Preserve sparse_embeddings in data_points_to_batch_update_records function (#721)
Added support for max_tokens attribute in ChatVertexAI (#714)
fix: Remove redundant debug messages (#713)
release 2.0.12 (#709)
fix tests
fix unit tests
vertexai: update standard tests library
fix linting (#708)
Fix additional_headers fallback in validate_parms_base (#700)
vertexai: relax pydantic version constraint (#696)
support codestral 25.01 (#695)
Add "mistral-large-2411@001" to model garden maas (#663)
release 2.0.11 (#692)
vertexai: Add Intro to the Vertex AI library (#660)
[VertexAI] Take vertexai init location as default location for BaseVertexAI (#683)
chore/update deps (#674)
added google_search_tool for gemini-2 (#677)
anthropic on vertexai: Update parsed parameter from image_url to image (#669)
vertexai: Add ChatAnthropicVertex Text Caching Support (#672)
[vertexai] Add Hybrid Search Capabilities to VertexAI Vector Search (#628)
vertexai: add probability_score and severity_score to safety_ratings (#656)
google-vertexai: release...

libs/genai/v3.0.0rc1

Initial release

increment versions
update deps
fix(genai): parse server tool call and result blocks (#1246)
Merge branch 'main' into v3.0.0
test: use gemini-flash-lite-latest (#1230)
chore(genai): bump version
release(genai): 3.0.0a1
bump genai lock/deps
feat(genai): (v1) server tool use (#1223)
feat(genai): standard content (#1204)
Merge branch 'main' into v3.0.0
chore(genai): update from gemini-1.5-flash to gemini-2.5-flash (#1221)
updates to _convert_to_parts
Merge branch 'main' into v3.0.0
fix(genai): Fix outdated doc for max_output_tokens default value (#1209)
Merge branch 'main' into v3.0.0
fix(genai): preserve field descriptions and enum values for Union types with None in function schemas (#1092)
chore: update projects/locks
Merge branch 'main' into v3.0.0
chore: bump lockfiles, resolve linting (genai, vertex, community) (#1203)
Merge branch 'main' into v3.0.0
refactor(genai): test_chat_models.py (#1202)
Merge branch 'main' into v3.0.0
feat(genai): handle all finish messages for AQA (#1201)
fix(genai): extract and include metadata (#941)
feat: return model_name and model_provider in response_metadata
chore: v1 migrations (#1195)
chore: update min vers in pyproject.toml files
chore: lower bound to Python 3.10, keep formatting py39
fix(genai): defer async client init for embeddings; avoid eager loop creation (#1183)
fix(genai): update json_mode to json_schema for structured output methods (#1188)
chore(genai): update model names in tests (#1187)
docs(genai): enhance structured output methods docstring (#1186)
test(genai): add test for ChatGoogleGenerativeAI class transport check (#831)
release(genai): 2.1.12 (#1181)
feat(genai): add timeout and max_retries handling in chat methods (#1180)
chore(genai): switch some model references in tests (#1178)
feat(genai): copy retry logic to async chat_with_retry (#1179)
refactor(genai): remove direct import of google.api_core and use specific exceptions (#1176)
feat(genai): add aliases for timeout and max_retries (#1175)
feat(genai): update model examples and enhance field descriptions in _BaseGoogleGenerativeAI (#1174)
genai, vertex: nits (#1173)
fix: use .invoke() instead of __call__ for LLMs (#1172)
release(genai): 2.1.11 (#1160)
Fix Gemini schema validation errors for Union and Optional[Dict] types (#1143)
genai: Fix: Include missing prompt_feedback in AIMessageChunk.response_metadata (#1104)
fix(genai): unrecognized enum value (#1158)
fix(genai): incorrect type annotation in ChatGoogleGenerativeAI._prepare_request (#1151)
genai: do not ignore Enum field definition (#1080)
feat: add VSCode settings for Ruff fixes (#1128)
fix(genai): Correct int/float type conversion in tool call args (#1110)
infra: poetry -> uv (#1101)
release(genai): 2.1.10 (#1095)
fix(genai): Update image input tests (#1094)
fix: Add field alias for default_metadata in ChatGoogleGenerativeAI (#1061)
fix: Update image URL in test for ChatGoogleGenerativeAI multimodal invocation (#1082)
genai: Enhance retry mechanism in ChatGoogleGenerativeAI with customizable parameter (#915)
[genai] docs: enhance multimodal input examples for ChatGoogleGenerativeAI (#1072)
release 2.1.9 (#1075)
vertexai/genai: Added missing chunk to on_llm_new_token to _stream_gemini (#1058)
fix: use new model names (#1067)
[genai, vertexai]: Add anyOf support for Union types in tool calling (#992)
fix: update thinking model name to match latest version (#1044)
fix(tests): embeddings and chat_models test fixes (#1035)
genai: release 2.1.8 (#1034)
fix for LangSmith
fix tuples
fix: ensure model prefix is applied consistently across classes // Remove tunedModel since it's no longer supported
fix docs
additional robustness for image generation tests
feat: add async embedding methods and update tests for output dimensionality for Gemini embedding compat
fix(docs): code blocks
various docs formatting fixes
genai: release 2.1.7
genai[patch]: fix totals in usage metadata and refactor (#1016)
genai: release 2.1.6 (#1012)
genai: Add json_mode support for structured output and schema reference handling in ChatGoogleGenerativeAI (#970)
genai[patch]: add test case for built-in search (#1009)
genai: fix multi-turn conversations with code execution blocks (#1003)
handle thinking blocks on input (#1007)
genai: Fix the issue task_type passed into embed_query is ignored. (#908)
genai: Added Audio Output support (#949)
1.5 -> 2.0 (#981)
add missing transport argument (#965)
genai: docs added complete and structured README for langchain-google-genai package (#933)
genai[patch]: support google_search and code_execution tools via dict (#958)
[genai]: Quick fix for "Unknown field for Part: thought" (#967)
bump core and increment version (#952)
add grounding_metadata support to ChatGoogleGenerativeAI (#944)
genai: Added include_thoughts support. (#937)
genai[patch]: use 1.5-latest in tests (#942)
fix: update model_fields access for Pydantic v2.11+ compatibility (#931)
genai[patch]: support image parts in tool responses (#921)
genai[patch]: fix prompt in flaky test and add retries (#901)
release 2.1.4 (#897)
fix(genai): Fix nested array schema conversion for function calling (#882)
[genai|vertexai|community]: Adding Vertex Tool Context into client info (#875)
fix: Trace model name without models/ prefix (#896)
genai: Added thinking_budget support to ChatGoogleGenerativeAI. (#884)
genai, vertexai[patch]: support model_kwargs (#891)
genai: fix issue where newlines were stripped during streaming (#894)
added model name to client-info (#885)
genai[patch]: add standard test for audio inputs (#869)
increment version (#865)
genai[patch]: support standard format for multi-modal content blocks (#862)
fix formatting (#850)
remove warning (#811)
removal of automatic 'models/' prefix addition for tuned type models (#842)
Docs: Add context caching examples (#835)
smol doc update (#833)
[genai, vertexai]: Add warnings for invalid kwargs (#829)
added check for chatgooglegenerativeai class (#822)
genai: Fix embed_query method to use the non-batch endpoint (#817)
bump core and increment version (#820)
genai[patch]: return model_name in response_metadata when streaming (#819)
readme: updated inverted commas for proper code block (#806)
(genai): Add support for Gemini code execution (#788)
genai: add grounding support for Gemini models (#795)
bump core and increment version (#800)
genai[patch]: update langchain-tests (#798)
genai: Added field response_modalities in ChatGoogleGenerativeAI initialization. (#792)
genai: Add image output parsing support to ChatGoogleGenerativeAI (#789)
fix[genai]: Allow Temperature Values Up to 2.0 for ChatGoogleGenerativeAI (#780)
release 2.0.12 (#776)
genai : filter out empty message contents in ChatGoogleGenerativeAI (#775)
genai[patch]: update langchain-tests, fix bugs with structured output (#772)
genai[patch]: update gemini 1.0 -> 1.5 and 2.0 in tests (#761)
release 2.0.11 (#745)
genai: Fix required field is ignored when dealing with nested pydantic models (#751)
Fix integration test with multiple parts (#681)
updating dependencies (#742)
genai[patch]: release 2.0.10 (#743)
genai: update standard tests library (#705)
genai: fix function_utils to parse object/array json schemas properly (#715)
fix: Fix tool calling with legacy agent (#712)
release 2.0.9 (#698)
update model (#684)
genai: Fix multiple tool calls in a single AIMessage (#671)
genai: Fixed nested pydantic structures recursion (#658)
langchain-google: remove possible trailing newline from content text in response candidates (#598)
google-genai: release 2.0.8 (#653)
enabled tool support for gemini2 (#640)
release 2.0.7 (#643)
Removed redundant lines that added the "title" field to the schema. (#634)
docs: fix google_genai typos (#614)
release genai 2.0.6 (#620)
genai[feat]: add filetype detection for images from bytes (#616)
genai: release 2.0.5
genai[refactoring]: Remove Pillow support, adjust dependencies, and clean up unused code (#603)
genai[patch]: fix tool call reading (#606)
langchain-standard-tests -> langchain-tests (#604)
genai: added usage metadata info in the llm results(#573) (#584)
genai[patch]: Release 2.0.4
genai[patch]: rm image file load (#585)
genai[patch]: rm image file load
google-genai: release 2.0.3 (#560)
Update _function_utils.py (#575)
fmt
genai[patch]: Release 2.0.2
genai[patch]: fix tool schema with nullable args (#566)
genai[patch]: trace tools in openai format (#565)
Added support for tools in flass series of models (#562)
genai: fix usage metadata (#545)
fmt
fmt
merge
genai[patch]: support caching and cached token counting
release (#543)
genai: fix pydantic structured_output with array (#469)
bump google-generativeai to 0.8.0 (#537)
fix standard_tests (#541)
genai[patch]: support max_tokens init arg (#517)
genai: check all BaseModel versions (#394)
genai: add integration_test help message for Makefile (#504)
genai: fix integration_test to avoid 429 Resource has been exhausted (#506)
community, genai, vertexai[major]: release 2.0 (#489)
remove inactive tests
genai[patch]: add standard tests (#478)
Revert "genai[patch]: add standard tests"
rate limit more
format
add standard tests
genai: use SecretStr for api_key (#397)
remove integration tests for bison (#477)
fix
nit
update genai
genai: add mkdir -p option for make lint (#470)
release (#461)
Add support for allOf in genai (#453)
merge
Merge branch 'main' into cc/llm_tracing_params
add genai
bump core (genai)
genai[patch]: Release 1.0.9
make ChatGoogleGenerativeAI serializable (#443)
genai[patch]: support media inputs (#442)
GenerateContentRequest.tools.function_declarations.parameters set None when function parameters is empty (#416)
genai: fix Makefile about ruff check (#419)
genai[patch]: with_structured_output support pydantic v2 (#414)
release (#388)
fix calling multiple tools + refactor to make closer to VertexAI impl… (#387)
format
update vision tests
vertexai[patch], genai[patch]: specify ToolCall "type" (#366)
Fix opti...

libs/genai/v3.0.0a1

Initial release

release(genai): 3.0.0a1
bump genai lock/deps
feat(genai): (v1) server tool use (#1223)
feat(genai): standard content (#1204)
Merge branch 'main' into v3.0.0
chore(genai): update from gemini-1.5-flash to gemini-2.5-flash (#1221)
updates to _convert_to_parts
Merge branch 'main' into v3.0.0
fix(genai): Fix outdated doc for max_output_tokens default value (#1209)
Merge branch 'main' into v3.0.0
fix(genai): preserve field descriptions and enum values for Union types with None in function schemas (#1092)
chore: update projects/locks
Merge branch 'main' into v3.0.0
chore: bump lockfiles, resolve linting (genai, vertex, community) (#1203)
Merge branch 'main' into v3.0.0
refactor(genai): test_chat_models.py (#1202)
Merge branch 'main' into v3.0.0
feat(genai): handle all finish messages for AQA (#1201)
fix(genai): extract and include metadata (#941)
feat: return model_name and model_provider in response_metadata
chore: v1 migrations (#1195)
chore: update min vers in pyproject.toml files
chore: lower bound to Python 3.10, keep formatting py39
fix(genai): defer async client init for embeddings; avoid eager loop creation (#1183)
fix(genai): update json_mode to json_schema for structured output methods (#1188)
chore(genai): update model names in tests (#1187)
docs(genai): enhance structured output methods docstring (#1186)
test(genai): add test for ChatGoogleGenerativeAI class transport check (#831)
release(genai): 2.1.12 (#1181)
feat(genai): add timeout and max_retries handling in chat methods (#1180)
chore(genai): switch some model references in tests (#1178)
feat(genai): copy retry logic to async chat_with_retry (#1179)
refactor(genai): remove direct import of google.api_core and use specific exceptions (#1176)
feat(genai): add aliases for timeout and max_retries (#1175)
feat(genai): update model examples and enhance field descriptions in _BaseGoogleGenerativeAI (#1174)
genai, vertex: nits (#1173)
fix: use .invoke() instead of __call__ for LLMs (#1172)
release(genai): 2.1.11 (#1160)
Fix Gemini schema validation errors for Union and Optional[Dict] types (#1143)
genai: Fix: Include missing prompt_feedback in AIMessageChunk.response_metadata (#1104)
fix(genai): unrecognized enum value (#1158)
fix(genai): incorrect type annotation in ChatGoogleGenerativeAI._prepare_request (#1151)
genai: do not ignore Enum field definition (#1080)
feat: add VSCode settings for Ruff fixes (#1128)
fix(genai): Correct int/float type conversion in tool call args (#1110)
infra: poetry -> uv (#1101)
release(genai): 2.1.10 (#1095)
fix(genai): Update image input tests (#1094)
fix: Add field alias for default_metadata in ChatGoogleGenerativeAI (#1061)
fix: Update image URL in test for ChatGoogleGenerativeAI multimodal invocation (#1082)
genai: Enhance retry mechanism in ChatGoogleGenerativeAI with customizable parameter (#915)
[genai] docs: enhance multimodal input examples for ChatGoogleGenerativeAI (#1072)
release 2.1.9 (#1075)
vertexai/genai: Added missing chunk to on_llm_new_token to _stream_gemini (#1058)
fix: use new model names (#1067)
[genai, vertexai]: Add anyOf support for Union types in tool calling (#992)
fix: update thinking model name to match latest version (#1044)
fix(tests): embeddings and chat_models test fixes (#1035)
genai: release 2.1.8 (#1034)
fix for LangSmith
fix tuples
fix: ensure model prefix is applied consistently across classes // Remove tunedModel since it's no longer supported
fix docs
additional robustness for image generation tests
feat: add async embedding methods and update tests for output dimensionality for Gemini embedding compat
fix(docs): code blocks
various docs formatting fixes
genai: release 2.1.7
genai[patch]: fix totals in usage metadata and refactor (#1016)
genai: release 2.1.6 (#1012)
genai: Add json_mode support for structured output and schema reference handling in ChatGoogleGenerativeAI (#970)
genai[patch]: add test case for built-in search (#1009)
genai: fix multi-turn conversations with code execution blocks (#1003)
handle thinking blocks on input (#1007)
genai: Fix the issue task_type passed into embed_query is ignored. (#908)
genai: Added Audio Output support (#949)
1.5 -> 2.0 (#981)
add missing transport argument (#965)
genai: docs added complete and structured README for langchain-google-genai package (#933)
genai[patch]: support google_search and code_execution tools via dict (#958)
[genai]: Quick fix for "Unknown field for Part: thought" (#967)
bump core and increment version (#952)
add grounding_metadata support to ChatGoogleGenerativeAI (#944)
genai: Added include_thoughts support. (#937)
genai[patch]: use 1.5-latest in tests (#942)
fix: update model_fields access for Pydantic v2.11+ compatibility (#931)
genai[patch]: support image parts in tool responses (#921)
genai[patch]: fix prompt in flaky test and add retries (#901)
release 2.1.4 (#897)
fix(genai): Fix nested array schema conversion for function calling (#882)
[genai|vertexai|community]: Adding Vertex Tool Context into client info (#875)
fix: Trace model name without models/ prefix (#896)
genai: Added thinking_budget support to ChatGoogleGenerativeAI. (#884)
genai, vertexai[patch]: support model_kwargs (#891)
genai: fix issue where newlines were stripped during streaming (#894)
added model name to client-info (#885)
genai[patch]: add standard test for audio inputs (#869)
increment version (#865)
genai[patch]: support standard format for multi-modal content blocks (#862)
fix formatting (#850)
remove warning (#811)
removal of automatic 'models/' prefix addition for tuned type models (#842)
Docs: Add context caching examples (#835)
smol doc update (#833)
[genai, vertexai]: Add warnings for invalid kwargs (#829)
added check for chatgooglegenerativeai class (#822)
genai: Fix embed_query method to use the non-batch endpoint (#817)
bump core and increment version (#820)
genai[patch]: return model_name in response_metadata when streaming (#819)
readme: updated inverted commas for proper code block (#806)
(genai): Add support for Gemini code execution (#788)
genai: add grounding support for Gemini models (#795)
bump core and increment version (#800)
genai[patch]: update langchain-tests (#798)
genai: Added field response_modalities in ChatGoogleGenerativeAI initialization. (#792)
genai: Add image output parsing support to ChatGoogleGenerativeAI (#789)
fix[genai]: Allow Temperature Values Up to 2.0 for ChatGoogleGenerativeAI (#780)
release 2.0.12 (#776)
genai : filter out empty message contents in ChatGoogleGenerativeAI (#775)
genai[patch]: update langchain-tests, fix bugs with structured output (#772)
genai[patch]: update gemini 1.0 -> 1.5 and 2.0 in tests (#761)
release 2.0.11 (#745)
genai: Fix required field is ignored when dealing with nested pydantic models (#751)
Fix integration test with multiple parts (#681)
updating dependencies (#742)
genai[patch]: release 2.0.10 (#743)
genai: update standard tests library (#705)
genai: fix function_utils to parse object/array json schemas properly (#715)
fix: Fix tool calling with legacy agent (#712)
release 2.0.9 (#698)
update model (#684)
genai: Fix multiple tool calls in a single AIMessage (#671)
genai: Fixed nested pydantic structures recursion (#658)
langchain-google: remove possible trailing newline from content text in response candidates (#598)
google-genai: release 2.0.8 (#653)
enabled tool support for gemini2 (#640)
release 2.0.7 (#643)
Removed redundant lines that added the "title" field to the schema. (#634)
docs: fix google_genai typos (#614)
release genai 2.0.6 (#620)
genai[feat]: add filetype detection for images from bytes (#616)
genai: release 2.0.5
genai[refactoring]: Remove Pillow support, adjust dependencies, and clean up unused code (#603)
genai[patch]: fix tool call reading (#606)
langchain-standard-tests -> langchain-tests (#604)
genai: added usage metadata info in the llm results(#573) (#584)
genai[patch]: Release 2.0.4
genai[patch]: rm image file load (#585)
genai[patch]: rm image file load
google-genai: release 2.0.3 (#560)
Update _function_utils.py (#575)
fmt
genai[patch]: Release 2.0.2
genai[patch]: fix tool schema with nullable args (#566)
genai[patch]: trace tools in openai format (#565)
Added support for tools in flass series of models (#562)
genai: fix usage metadata (#545)
fmt
fmt
merge
genai[patch]: support caching and cached token counting
release (#543)
genai: fix pydantic structured_output with array (#469)
bump google-generativeai to 0.8.0 (#537)
fix standard_tests (#541)
genai[patch]: support max_tokens init arg (#517)
genai: check all BaseModel versions (#394)
genai: add integration_test help message for Makefile (#504)
genai: fix integration_test to avoid 429 Resource has been exhausted (#506)
community, genai, vertexai[major]: release 2.0 (#489)
remove inactive tests
genai[patch]: add standard tests (#478)
Revert "genai[patch]: add standard tests"
rate limit more
format
add standard tests
genai: use SecretStr for api_key (#397)
remove integration tests for bison (#477)
fix
nit
update genai
genai: add mkdir -p option for make lint (#470)
release (#461)
Add support for allOf in genai (#453)
merge
Merge branch 'main' into cc/llm_tracing_params
add genai
bump core (genai)
genai[patch]: Release 1.0.9
make ChatGoogleGenerativeAI serializable (#443)
genai[patch]: support media inputs (#442)
GenerateContentRequest.tools.function_declarations.parameters set None when function parameters is empty (#416)
genai: fix Makefile about ruff check (#419)
genai[patch]: with_structured_output support pydantic v2 (#414)
release (#388)
fix calling multiple tools + refactor to make closer to VertexAI impl… (#387)
format
update vision tests
vertexai[patch], genai[patch]: specify ToolCall "type" (#366)
Fix optional fields in Pydantic models (#363)
fix async calls without event_loop (#375)
vertexai[patch], genai[patch]: use tool name explicitly in with struc… (#341)
genai[patch]: fix tool calling, add do...

libs/genai/v2.1.12

Changes since libs/genai/v2.1.11

release(genai): 2.1.12 (#1181)
feat(genai): add timeout and max_retries handling in chat methods (#1180)
chore(genai): switch some model references in tests (#1178)
feat(genai): copy retry logic to async chat_with_retry (#1179)
refactor(genai): remove direct import of google.api_core and use specific exceptions (#1176)
feat(genai): add aliases for timeout and max_retries (#1175)
feat(genai): update model examples and enhance field descriptions in _BaseGoogleGenerativeAI (#1174)
genai, vertex: nits (#1173)
fix: use .invoke() instead of __call__ for LLMs (#1172)

libs/vertexai/v2.1.2

Changes since libs/vertexai/v2.1.1

release(vertexai): 2.1.2 (#1161)
refactor(vertex): resolve test warnings (#1166)
fix(vertexai): handle unrecognized FinishReason values (#1164)
fix(vertexai): update audio timestamp test regex pattern (#1163)
chore(vertex): update embedding models to use 'gemini-embedding-001' and 'imagen-3.0' (#1086)
fix(vertex): use imagen-3.0-generate-002 as default model for image generation (#1162)
VectorSearchVectorStore: integer handling in numeric restrictions (#1155)
Guard against unknown FinishReason values (#1085)
fix(vertexai): add Union type support to V1 schema formatter (#1137)
vertex: add http_client support to ChatAnthropicVertex (#1154)

libs/community/v2.0.10

Changes since libs/community/v2.0.9

release 2.0.10 (#1168)
community: added model armor sanitization runnables (#982)

libs/vertexai/v2.1.1

Changes since libs/vertexai/v2.1.0

release(vertexai): 2.1.1 (#1149)
vertex: fix AIMessage tool call handling with empty content (#1145)
vertexai(patch): fix function calling for v1 endpoints (#1147)
vertexai: handle messages that evaluate to a literal float (#938)

libs/genai/v2.1.11

Changes since libs/genai/v2.1.10

release(genai): 2.1.11 (#1160)
Fix Gemini schema validation errors for Union and Optional[Dict] types (#1143)
genai: Fix: Include missing prompt_feedback in AIMessageChunk.response_metadata (#1104)
fix(genai): unrecognized enum value (#1158)
fix(genai): incorrect type annotation in ChatGoogleGenerativeAI._prepare_request (#1151)
genai: do not ignore Enum field definition (#1080)
feat: add VSCode settings for Ruff fixes (#1128)
fix(genai): Correct int/float type conversion in tool call args (#1110)
infra: poetry -> uv (#1101)

libs/community/v2.0.9

Changes since libs/community/v2.0.8

release(community): 2.0.9 (#1159)
community: fix operator precedence issue in credential handling (#1152)
Add test coverage for optional BigQuery dependency (#1027)
community: fix CalendarUpdateEvent schema-implementation mismatch for optional parameters (#1153)

libs/vertexai/v2.1.0

What's Changed

• infra: poetry -> uv by @mdrxy in #1101
• fix(genai): Correct int/float type conversion in tool call args by @codeonbush in #1110
• vertexai: resolve errors from uv migration by @ccurme in #1125
• fix(vertexai): resolve Pydantic deprecation warning in VertexAI LLM by @mdrxy in #1127
• vertexai: require opt-in for ast.literal_eval on input message strings by @ccurme in #1122
• test(vertexai): suppress UserWarning in parameter tests by @mdrxy in #1130
• test(vertexai): add pytest warnings filter for unsupported parameters in ChatVertexAI by @mdrxy in #1129
• feat: add VSCode settings for Ruff fixes by @mdrxy in #1128
• vertex: release 2.1.0 by @ccurme in #1132
• fix(vertexai): Set requires_reference=True for VertexPairWiseStringEvaluator by @mdrxy in #1133
• chore(tests): add vertex type ignore specifiers by @mdrxy in #1134
• infra: fix release workflow by @ccurme in #1135
• infra: fix release workflow by @ccurme in #1136
• vertexai: handle messages that evaluate to a literal float by @JoshuaPostel in #938

New Contributors

• @codeonbush made their first contribution in #1110
• @JoshuaPostel made their first contribution in #938

Full Changelog: libs/genai/v2.1.10...libs/vertexai/v2.1.0