LangChain

This patch release for langchain-anthropic primarily fixes an issue by normalizing cross-provider tool-call IDs and includes several dependency bumps.

This release primarily updates the underlying perplexityai dependency to version 0.34.1.

This release primarily addresses a serialization bug within the fireworks integration by ensuring non-wire keys are stripped from content parts. It also updates internal dependencies.

Version 1.3.0 introduces the `use_responses_api` flag for ChatPerplexity and updates several internal dependencies.

This release introduces PII redaction and stream transformer registration features, alongside fixes for TodoListMiddleware and glob_search sorting. It also updates several internal dependencies.

This release focuses on maintenance, dependency bumps, and minor fixes within the OpenAI integration, including guarding httpx finalizers and correctly sourcing LLM context sizes.

This release bumps the idna dependency and introduces an enhancement to streaming assertions within the standard-tests library.

Version 1.4.1 for langchain-fireworks introduces improved resilience by adding retries for API connection errors and setting a default retry limit.

Version 1.4.0 for langchain-fireworks migrates the integration to the fireworks-ai 1.x SDK and introduces proper context overflow error handling. Several minor dependency bumps were also performed.

This release primarily consists of dependency bumps and maintenance updates, including refreshing stale OpenAI model references and fixing a test assertion regarding model overrides.

This patch release (1.3.1) addresses a bug related to Bedrock provider aliasing within the summarization token check.

This release introduces support for version="v3" in event streaming methods for langchain agents.

This release focuses heavily on infrastructure updates, dependency bumps, and significant hardening across security (SSRF) and tracing mechanisms. New features include content-block-centric streaming (v2) and enhanced metadata tracing for LLM invocations.

This release reverts a previous change regarding agent tagging and introduces deprecations for 'hub' and load/dump limits within langchain-classic.

This release backports a critical path-traversal security fix to the langchain-core v0.3 branch.

This release backports security hardening for loads/dumps to version 0.3 and deprecates the 'hub' functionality.

This release of langchain-classic deprecates the 'hub' functionality and limits loads/dumps operations.

This release introduces significant updates to streaming protocols (v2 and v3), enhances agent capabilities with new middleware features, and includes numerous dependency bumps and bug fixes across various components.

This release primarily focuses on security hardening for manifest loading and restricts deserialization in the classic storage module, while also updating a dependency.

This release focuses on security enhancements by restricting deserialization and hardening the `load()` function against untrusted manifests.

This release focuses on security by hardening the `load()` function against untrusted manifests.

This release (1.3.3) focuses on hardening security for manifest loading and fixing several internal logic issues, including batch validation and tracer data preservation.

Version 1.3.1 of langchain-fireworks includes fixes related to API key requirements in embeddings and cleaning up ToolMessage content blocks.

This release primarily addresses a bug fix in the mistralai integration by cleaning up keys in ToolMessage.

This release primarily focuses on refactoring deprecation targets within langchain-classic and includes a minor bug fix for httpx finalizers in the anthropic integration.

This release focuses on dependency updates, minor bug fixes in Fireworks, and introduces content-block-centric streaming (v2) in the core library. Deprecation messages have also been refined.

This release (0.2.3) for langchain-openrouter primarily addresses a streaming bug related to reasoning details.

This release for langchain-mistralai adds image input support for human messages and includes several dependency updates and infrastructure improvements across the ecosystem.

The Fireworks integration has been updated to version 1.3.0, introducing the ability to specify a `service_tier` when initializing `ChatFireworks`.

This release includes numerous performance improvements, dependency bumps, and feature additions like content-block-centric streaming (v2) and integration of stream_events(version='v3'). Several bug fixes address token counting, model recognition, and middleware configuration.

This release introduces new fields for session tracking in OpenRouter and implements content-block-centric streaming (v2). It also includes several dependency updates and a fix for usage metadata extraction across multiple providers.

This release introduces content-block-centric streaming (v2) and support for the v3 stream protocol, alongside numerous bug fixes related to tracing, validation, and dependency updates. Path saving/loading functions for prompts have been deprecated.

This release introduces significant feature updates, including content-block-centric streaming (v2) and support for the v3 event protocol. Numerous bug fixes address issues related to tracing, serialization, SSRF hardening, and model wrapper compatibility.

Version 1.2.1 primarily addresses minor dependency updates and includes a fix for multimodal content block translation in Fireworks, alongside a new content-block-centric streaming feature in core.

This release introduces a new decision type to the HITL middleware for enhanced control flow.

This release introduces `PerplexityEmbeddings`, overhauls the Perplexity integration with its official SDK, and adds new fields to model profiles. It also includes several dependency bumps, notably for security and stability.

This release introduces content-block-centric streaming (v2) in core and adds agent type tagging for `create_agent` calls. It also includes several dependency updates and performance improvements.

This release (1.1.7) introduces content-block-centric streaming (v2) and updates several dependencies, including bumping the minimum required versions for core components.

This release for langchain-anthropic (1.4.2) restores cache control functionality and introduces content-block-centric streaming (v2) in the core library, alongside bumping minimum dependency versions.

This hotfix release bumps minimum core dependency versions and introduces content-block-centric streaming (v2) while fixing an issue with the Responses API check.

This release introduces content-block-centric streaming (v2) to the core library.

This release updates dependency versions across the board, refreshes model profile data, and introduces streaming usage metadata support for the Fireworks integration. It also fixes several bugs related to retries and tool binding in Fireworks.

This release updates dependencies and introduces minor feature enhancements in the core library, including changes to output formatting and tracer metadata handling.

Version 1.2.0 introduces a fix to prevent silent streaming hangs within ChatOpenAI and includes minor CI and partner integration chore updates.

This release primarily addresses a bug related to prompt cache retention during streaming operations in langchain-openai.

This release of langchain-openai primarily focuses on bug fixes related to streaming responses and Azure profile inference.

Langchain-core version 1.3.0 introduces new traceable metadata fields for chat models and LLMs, alongside several bug fixes related to SSRF policies and garbage collection support.

This release (1.4.1) introduces support for new Anthropic model features like Opus 4.7 and adaptive thinking, alongside dependency updates and minor bug fixes.

This release focuses on hardening HuggingFace integrations by improving repository ID validation and optimizing API calls, alongside updating model profile data and dependencies.

This release introduces new features like enhanced metadata tracing, support for ChatBaseten, and multimodal token counting. It also includes numerous bug fixes related to SSRF hardening, model serialization, and tool handling, alongside the deprecation of prompt saving/loading functions due to added path validation.

This release of langchain-classic (1.0.4) introduces deprecations, dependency bumps including pytest and cryptography, and fixes related to mypy errors and Azure AI Foundry model providers.

Version 1.1.14 of langchain-openai fixes an issue related to image token counting by implementing SSRF-safe transport. This release also updates several dependencies.

This release updates numerous dependencies, enforces SSRF-safe transport for URL splitting, and fixes a data loss issue in RecursiveJsonSplitter.

This release ports change 36816 to the v1.2 branch, resulting in version 1.2.31 of langchain-core.

This release primarily focuses on hardening private SSRF utilities within the core library.

This release primarily focuses on bug fixes related to API response handling, token extraction, and User-Agent overrides, alongside adding a feature to impute placeholder filenames for OpenAI file inputs.

This release updates langchain-core to version 1.2.29 and incorporates a fix from PR #36725.

This release focuses heavily on stability, security hardening (anti-SSRF), and improved token counting, especially for multimodal inputs and tool schemas. Several prompt loading/saving methods have been deprecated due to path validation improvements.

This release focuses heavily on stability, security hardening (anti-SSRF, pygments update), and feature parity, including adding Bedrock and OpenRouter support, and enhancing token counting for multimodal inputs and tool schemas. Several internal cleanups and deprecations related to prompt saving paths were also introduced.

This release primarily focuses on improving prompt sanitization within the core library, moving from version 0.3.83 to 0.3.84.

This release primarily focuses on improving template security by adding more sanitization.

This release primarily focuses on dependency updates, security patches (pygments), and minor fixes to standard tests for sandbox backends.

This release primarily addresses a bug related to handling symlinks in the deprecated prompt saving path and updates a dependency comment.

This release introduces new features for Ollama, including support for `response_format`, `dimensions` in embeddings, and logprobs, alongside several bug fixes and dependency updates.

Version 1.2.26 introduces a new serializable mapping for ChatBaseten and fixes serialization/validation issues for Bedrock models.

This release primarily updates the dependency version for aiohttp within the langchain_v1 library.

This release (1.2.25) focuses on minor fixes, including hardening file checks in deprecated prompt loading and correcting documentation typos.

Version 1.2.24 introduces placeholder filename imputation for OpenAI file inputs and updates the dependency constraint for Pygments.

This release focuses on dependency updates, performance improvements, and several bug fixes, including updates for Azure AI Foundry providers and recursion limit handling in agents.

This release bumps the dependency on 'requests' and fixes an issue in the OpenRouter integration where attribution headers were not being passed correctly.

This release reverts a previous fix related to tracing invocation parameters and updates the 'requests' dependency version.

This release updates the default search type in langchain-exa to 'auto' and includes numerous dependency bumps across the project.

The OpenRouter integration received an update to include marketplace attribution fields, and core components received fixes for missing model profile fields.

Version 1.2.22 introduces path validation for prompt saving and loading, leading to the deprecation of the underlying methods.

This release (1.1.12) focuses on bug fixes for OpenAI integrations, including support for the phase parameter and fixing file descriptor leaks, alongside updates to model profiles.

This release primarily addresses a bug fix in model profiles by adding missing fields and implementing schema drift warnings, alongside minor cleanup.

This release includes minor dependency bumps, CI improvements, and adds LangSmith integration metadata to agent and chat model initialization functions. A bug fix was also applied to the OpenAI Responses API input typing.

Version 1.2.20 of langchain-core includes security hardening, improved LangSmith integration metadata, and a fix for tracing invocation parameters.

Version 1.4 introduces explicit caching for system messages and tool definitions via AnthropicPromptCachingMiddleware and improves cache control delegation.

This release (1.3.5) for langchain-anthropic focuses on bug fixes related to caching, streaming, and tool choice handling, alongside dependency updates and new model profile features.

This release introduces new fields to model profiles and fixes several bugs related to exception handling and type resolution. Dependencies, including langsmith and urllib3, have also been updated.

This release primarily involves internal housekeeping, moving BaseCrossEncoder to langchain-core and updating several minor and patch dependencies across the project.

This release moves the BaseCrossEncoder class to langchain-core and updates the tornado dependency version.

This release introduces tracing capabilities for wrapped models and tool calls within langchain_v1.

This release introduces support for OpenRouter and automatic server-side compaction in OpenAI, alongside several dependency updates and bug fixes related to model initialization and detector output.

This release introduces support for tool search and streaming token usage for OpenRouter, alongside several bug fixes related to model detection and structured output handling. It also bumps minimum dependency versions.

This release (1.2.18) focuses on minor fixes, including correcting docstrings and preserving schema factory settings, alongside adding tool search support for OpenAI.

This release bumps the minimum required version for `langchain-core` to 0.3.73 and patches a critical ReDoS vulnerability in MRKL/ReAct regexes. It also includes numerous internal cleanups, dependency bumps (like Ruff), and updates to model handling, particularly for Anthropic and Google models.

This release bumps `langchain-classic` to 1.0.2, patching a ReDoS vulnerability and including various dependency updates. New features include support for automatic server-side compaction in OpenAI integrations.

This release bumps the version of langchain-openrouter to 0.1.0 and introduces several new features, including streaming token usage and cost reporting for OpenRouter, alongside various infrastructure and profile updates.

Version 1.2.17 of langchain-core includes a fix for extracting usage metadata from tracer messages and several dependency bumps.

This release focuses on dependency updates, resolving compatibility issues with huggingface-hub 1.x, and introducing new fields to model profiles.

This release of langchain-core includes a minor bug fix related to handling empty tool chunk IDs during merging.

This release focuses on improving the Anthropic integration by adding Bedrock support and a User-Agent header, alongside several fixes related to model IDs and response handling.

This release focuses on bug fixes in the core library, improved error handling, and the addition (and subsequent reversion) of the ChatAnthropicBedrock wrapper. Performance improvements were also made by deferring specific langsmith imports.

This release focuses primarily on bug fixes within the core library, addressing issues related to tool call merging, recursion errors, and LangSmith tracing parameter handling. Dependency groups were also updated.

Version 1.1.1 of langchain-text-splitters focuses on bug fixes, including resolving an SSRF vulnerability and mutation issues in splitters, alongside adding support for model_kwargs in SentenceTransformersTokenTextSplitter.

This release bumps several internal dependency groups, most notably updating langsmith to version 0.6.3, and releases version 1.1.5 of langchain-tests.

This release introduces support for automatic server-side compaction in OpenAI and adds the OpenRouter provider package. It also includes several bug fixes related to model properties, error handling, and dependency version bumps.

This release primarily involves bumping the core version and silencing a warning within the langchain-openrouter integration.