v0.5.2

What's Changed

• Prep for v0.5.1 by @brainlid in #454
• fix(ChatVertexAI): Fix tool calls for ChatVertexAI and support for Gemini 3 models by @abh1shek-sh in #452
• Revising tool detection by @brainlid in #458
• prep for v0.5.2 release by @brainlid in #459

Full Changelog: v0.5.1...v0.5.2

v0.5.1

What's Changed

• fix(ChatMistralAI): Handle broken tool names that kill LLMChain by @xu-chris in #448
• Handle thought content parts from Google Vertex by @mattmatters in #430
• feat: Verify function parameters before executing, reject faulty message deltas by @xu-chris in #449
• fix(ChatOpenAIResponses): Handle failed response status from OpenAI Responses API by @xu-chris in #450
• Expanding callbacks for tool detection and UI feedback by @brainlid in #453

Full Changelog: v0.5.0...v0.5.1

v0.5.0

What's Changed

• Add req_config to ChatOpenAIResponses by @xxdavid in #415
• Add thinking config to vertex ai by @mattmatters in #423
• Add support for OpenAI Response API Stateful context by @cjimison in #425
• Fixes image file_id content type for ChatOpenAIResponses by @sezaru in #438
• fix(ChatMistralAI): missing error handling and fallback mechanism on server outages (#434) by @xu-chris in #435
• feat(ChatMistralAI): add support for parallel tool calls by @xu-chris in #433
• use elixir 1.17 by @nbw in #427
• feat(GoogleChatAI): add thought_signature support for Gemini 3 function calls by @abh1shek-sh in #431
• Don't include top_p for gpt-5.2+ in ChatOpenAIResponses by @montebrown in #428
• Fix Mistral 'thinking' content parts by @arjan in #418
• Add verbose_api field to ChatPerplexity and ChatMistralAI by @arjan in #416
• Add support for OpenAI reasoning/thinking events in ChatOpenAIResponses by @arjan in #421
• Add new reasoning effort values to ChatOpenAIResponses by @xxdavid in #419
• fix "Support reasoning_content of deepseek model" introducing UI bug … by @kayuapi in #429
• Support json schema in vertex ai by @mattmatters in #424
• Base work for new agent library by @brainlid in #442
• fix ChatGrok tool call arguments and message flattening by @KristerV in #420
• Revert "fix ChatGrok tool call arguments and message flattening" by @brainlid in #445
• prep for v0.5.0 release by @brainlid in #451

New Contributors

• @cjimison made their first contribution in #425
• @sezaru made their first contribution in #438
• @xu-chris made their first contribution in #435
• @nbw made their first contribution in #427
• @abh1shek-sh made their first contribution in #431
• @kayuapi made their first contribution in #429
• @KristerV made their first contribution in #420

Full Changelog: v0.4.1...v0.5.0

v0.4.1

What's Changed

• OpenAI responses API improvements by @arjan in #391
• Support Anthropic disable_parallel_tool_use tool_choice setting by @vlymar in #390
• Update gettext dependency version to 1.0 by @bijanbwb in #393
• Add DeepSeek chat model integration by @gilbertwong96 in #394
• Loosen the gettext dependency by @montebrown in #399
• add MessageDelta.merge_deltas/2 by @brainlid in #401
• formatting update by @brainlid in #402
• Added an :cache_messages option for ChatAnthropic, can improve cache utilization. by @montebrown in #398
• Add support for Anthropic API PDF reading to the ChatAnthropic model. by @jadengis in #403
• feat: Add support for :file_url to ChatAnthropic too by @jadengis in #404
• Support reasoning_content of deepseek model by @gilbertwong96 in #407
• Add req_opts to ChatAnthropic by @stevehodgkiss in #408
• Open AI Responses API: Add support for file_url with link to file by @reetou in #395
• Add strict tool use support to ChatAnthropic by @stevehodgkiss in #409
• Add strict to function of ChatModels.ChatOpenAI by @nallwhy in #301
• Allow multi-part tool responses. by @montebrown in #410
• prep for v0.4.1 release by @brainlid in #411

New Contributors

• @vlymar made their first contribution in #390
• @bijanbwb made their first contribution in #393
• @gilbertwong96 made their first contribution in #394
• @reetou made their first contribution in #395

Full Changelog: v0.4.0...v0.4.1

v0.4.0

What's Changed since v0.3.3

• Add OpenAI and Claude thinking support - v0.4.0-rc.0 by @brainlid in #297
• vertex ai file url support by @ahsandar in #296
• Update docs for Vertex AI by @ahsandar in #304
• Fix ContentPart migration by @mathieuripert in #309
• Fix tests for content_part_for_api/2 of ChatOpenAI in v0.4.0-rc0 by @nallwhy in #300
• Fix tool_calls nil messages by @udoschneider in #314
• feat: Add structured output support to ChatMistralAI by @mathieuripert in #312
• feat: add configurable tokenizer to text splitters by @mathieuripert in #310
• simple formatting issue by @Bodhert in #307
• Update Message.new_system spec to accurately accept [ContentPart.t()]… by @rtorresware in #315
• Fix: Add token usage to ChatGoogleAI message metadata by @mathieuripert in #316
• feat: include raw API responses in LLM error objects for better debug… by @TwistingTwists in #317
• expanded docs and test coverage for prompt caching by @brainlid in #325
• Fix AWS Bedrock stream decoder ordering issue by @stevehodgkiss in #327
• significant updates for v0.4.0-rc.1 by @brainlid in #328
• filter out empty lists in message responses by @brainlid in #333
• fix: Require gettext ~> 0.26 by @mweidner037 in #332
• Add retry: transient to Req for Anthropic models in stream mode by @jonator in #329
• fixed issue with poorly matching list in case by @brainlid in #334
• feat: Add organization ID as a parameter by @hjemmel in #337
• Add missing verbose_api field to ChatOllamaAI for streaming compatibility by @gur-xyz in #341
• Added usage data to the VertexAI Message response. by @raulchedrese in #335
• feat: add run mode: step by @CaiqueMitsuoka in #343
• feat: add support for multiple tools in run_until_tool_used by @fortmarek in #345
• Fix ChatOllamaAI stop sequences: change from string to array type by @gur-xyz in #342
• expanded logging for ChatAnthropic API errors by @brainlid in #349
• Prevent crash when ToolResult with string in ChatGoogleAI.for_api/1 by @nallwhy in #352
• Bedrock OpenAI-compatible API compatibility fix by @stevehodgkiss in #356
• added xAI Grok chat model support by @alexfilatov in #338
• Support thinking to ChatGoogleAI by @nallwhy in #354
• Add req_config to ChatMode.ChatGoogleAI by @nallwhy in #357
• Clean up treating MessageDelta in ChatModels.ChatGoogleAI by @nallwhy in #353
• Expose full response headers through a new on_llm_response_headers callback by @brainlid in #358
• only include "user" with OpenAI request when a value is provided by @brainlid in #364
• Handle no content parts responses in ChatGoogleAI by @nallwhy in #365
• Adds support for gpt-image-1 in LangChain.Images.OpenAIImage by @Ven109 in #360
• Pref for release v0.4.0-rc.2 by @brainlid in #366
• fix: handle missing finish_reason in streaming responses for LiteLLM compatibility by @fbettag in #367
• Add support for native tool calls to ChatVertexAI by @raulchedrese in #359
• Adds should_continue? optional function to mode step by @CaiqueMitsuoka in #361
• Add OpenAI Deep Research integration by @fbettag in #336
• Add parallel_tool_calls option to ChatOpenAI model by @martosaur in #371
• Add optional AWS session token handling in BedrockHelpers by @quangngd in #372
• fix: handle LiteLLM responses with null b64_json in OpenAIImage by @fbettag in #368
• Add Orq AI chat by @arjan in #377
• Add req_config to ChatModels.ChatOpenAI by @koszta in #376
• fix(ChatGoogleAI): Handle cumulative token usage by @mweidner037 in #373
• fix(ChatGoogleAI): Prevent error from thinking content parts by @mweidner037 in #374
• feat(ChatGoogleAI): Full thinking config by @mweidner037 in #375
• Support verbosity parameter for ChatOpenAI by @rohan-b99 in #379
• add retry_on_fallback? to chat model definition and all models by @brainlid in #350
• Prep for v0.4.o-rc.3 by @brainlid in #380
• Use moduledoc instead of doc for LLMChain documentation by @xxdavid in #384
• Support OTP 28 in CI by @kianmeng in #382
• OpenAI responses by @vasspilka in #381
• Add AGENTS.md and CLAUDE.md file support by @brainlid in #385
• Suppress the compiler warning messages for ChatBumblebee by @brainlid in #386
• fix: Support for json-schema in OpenAI responses API by @vasspilka in #387
• Prepare for v0.4.0 release by @brainlid in #388

New Contributors

• @ahsandar made their first contribution in #296
• @mathieuripert made their first contribution in #309
• @udoschneider made their first contribution in #314
• @Bodhert made their first contribution in #307
• @rtorresware made their first contribution in #315
• @TwistingTwists made their first contribution in #317
• @mweidner037 made their first contribution in #332
• @jonator made their first contribution in #329
• @hjemmel made their first contribution in #337
• @gur-xyz made their first contribution in #341
• @CaiqueMitsuoka made their first contribution in #343
• @fortmarek made their first contribution in #345
• @alexfilatov made their first contribution in #338
• @Ven109 made their first contribution in #360
• @martosaur made their first contribution in #371
• @quangngd made their first contribution in #372
• @arjan made their first contribution in #377
• @koszta made their first contribution in #376
• @rohan-b99 made their first contribution in #379
• @xxdavid made their first contribution in #384

Full Changelog: v0.3.3...v0.4.0

v0.4.0-rc.3

What's Changed

• fix: handle missing finish_reason in streaming responses for LiteLLM compatibility by @fbettag in #367
• Add support for native tool calls to ChatVertexAI by @raulchedrese in #359
• Adds should_continue? optional function to mode step by @CaiqueMitsuoka in #361
• Add OpenAI Deep Research integration by @fbettag in #336
• Add parallel_tool_calls option to ChatOpenAI model by @martosaur in #371
• Add optional AWS session token handling in BedrockHelpers by @quangngd in #372
• fix: handle LiteLLM responses with null b64_json in OpenAIImage by @fbettag in #368
• Add Orq AI chat by @arjan in #377
• Add req_config to ChatModels.ChatOpenAI by @koszta in #376
• fix(ChatGoogleAI): Handle cumulative token usage by @mweidner037 in #373
• fix(ChatGoogleAI): Prevent error from thinking content parts by @mweidner037 in #374
• feat(ChatGoogleAI): Full thinking config by @mweidner037 in #375
• Support verbosity parameter for ChatOpenAI by @rohan-b99 in #379
• add retry_on_fallback? to chat model definition and all models by @brainlid in #350
• Prep for v0.4.o-rc.3 by @brainlid in #380

New Contributors

• @martosaur made their first contribution in #371
• @quangngd made their first contribution in #372
• @arjan made their first contribution in #377
• @koszta made their first contribution in #376
• @rohan-b99 made their first contribution in #379

Full Changelog: v0.4.0-rc.2...v0.4.0-rc.3

v0.4.0-rc.2

What's Changed

• filter out empty lists in message responses by @brainlid in #333
• fix: Require gettext ~> 0.26 by @mweidner037 in #332
• Add retry: transient to Req for Anthropic models in stream mode by @jonator in #329
• fixed issue with poorly matching list in case by @brainlid in #334
• feat: Add organization ID as a parameter by @hjemmel in #337
• Add missing verbose_api field to ChatOllamaAI for streaming compatibility by @gur-xyz in #341
• Added usage data to the VertexAI Message response. by @raulchedrese in #335
• feat: add run mode: step by @CaiqueMitsuoka in #343
• feat: add support for multiple tools in run_until_tool_used by @fortmarek in #345
• Fix ChatOllamaAI stop sequences: change from string to array type by @gur-xyz in #342
• expanded logging for ChatAnthropic API errors by @brainlid in #349
• Prevent crash when ToolResult with string in ChatGoogleAI.for_api/1 by @nallwhy in #352
• Bedrock OpenAI-compatible API compatibility fix by @stevehodgkiss in #356
• added xAI Grok chat model support by @alexfilatov in #338
• Support thinking to ChatGoogleAI by @nallwhy in #354
• Add req_config to ChatMode.ChatGoogleAI by @nallwhy in #357
• Clean up treating MessageDelta in ChatModels.ChatGoogleAI by @nallwhy in #353
• Expose full response headers through a new on_llm_response_headers callback by @brainlid in #358
• only include "user" with OpenAI request when a value is provided by @brainlid in #364
• Handle no content parts responses in ChatGoogleAI by @nallwhy in #365
• Adds support for gpt-image-1 in LangChain.Images.OpenAIImage by @Ven109 in #360
• Pref for release v0.4.0-rc.2 by @brainlid in #366

New Contributors

• @mweidner037 made their first contribution in #332
• @jonator made their first contribution in #329
• @hjemmel made their first contribution in #337
• @gur-xyz made their first contribution in #341
• @CaiqueMitsuoka made their first contribution in #343
• @fortmarek made their first contribution in #345
• @alexfilatov made their first contribution in #338
• @Ven109 made their first contribution in #360

Full Changelog: v0.4.0-rc.1...v0.4.0-rc.2

v0.4.0-rc.1

Refer to the CHANGELOG.md for notes on breaking changes and migrating.

What's Changed

• vertex ai file url support by @ahsandar in #296
• Update docs for Vertex AI by @ahsandar in #304
• Fix ContentPart migration by @mathieuripert in #309
• Fix tests for content_part_for_api/2 of ChatOpenAI in v0.4.0-rc0 by @nallwhy in #300
• Fix tool_calls nil messages by @udoschneider in #314
• feat: Add structured output support to ChatMistralAI by @mathieuripert in #312
• feat: add configurable tokenizer to text splitters by @mathieuripert in #310
• simple formatting issue by @Bodhert in #307
• Update Message.new_system spec to accurately accept [ContentPart.t()]… by @rtorresware in #315
• Fix: Add token usage to ChatGoogleAI message metadata by @mathieuripert in #316
• feat: include raw API responses in LLM error objects for better debug… by @TwistingTwists in #317
• expanded docs and test coverage for prompt caching by @brainlid in #325
• Fix AWS Bedrock stream decoder ordering issue by @stevehodgkiss in #327
• significant updates for v0.4.0-rc.1 by @brainlid in #328

New Contributors

• @ahsandar made their first contribution in #296
• @mathieuripert made their first contribution in #309
• @udoschneider made their first contribution in #314
• @Bodhert made their first contribution in #307
• @rtorresware made their first contribution in #315
• @TwistingTwists made their first contribution in #317

Full Changelog: v0.4.0-rc.0...v0.4.0-rc.1

v0.4.0-rc.0

What's Changed

• Add OpenAI and Claude thinking support - v0.4.0-rc.0 by @brainlid in #297

Introduces breaking changes while adding expanded support for thinking models.

NOTE: See the CHANGELOG.md for more details
IMPORTANT: Not all models are supported with this RC.

Full Changelog: v0.3.3...v0.4.0-rc.0

v0.3.3

What's Changed

• upgrade gettext and migrate by @brainlid in #271
• Support caching tool results for Anthropic calls by @ci in #269
• Fix OpenAI verbose_api by @aaparmeggiani in #274
• Support choice of Anthropic beta headers by @ci in #273
• Fix specifying media uris for google vertex by @mattmatters in #242
• feat: add support for pdf content with OpenAI model by @bwan-nan in #275
• feat: File urls for Google by @vasspilka in #286
• support streaming responses from mistral by @manukall in #287
• Support for json_response in ChatModels.ChatGoogleAI by @nallwhy in #277
• Fix options being passed to the ollama chat api by @alappe in #179
• Support for file with file_id in ChatOpenAI by @nallwhy in #283
• added LLMChain.run_until_tool_used/3 by @brainlid in #292
• adds telemetry by @epinault in #284

New Contributors

• @ci made their first contribution in #269
• @aaparmeggiani made their first contribution in #274
• @mattmatters made their first contribution in #242
• @vasspilka made their first contribution in #286
• @manukall made their first contribution in #287
• @epinault made their first contribution in #284

Full Changelog: v0.3.2...v0.3.3