Local / Open Source
LangChain
Updates
- Feature2d ago· GitHub Releases
langchain-openai==1.6.7
langchain-openai 1.6.7 adds Azure workload identity discovery, refreshes model profile data, and removes live tests for retired completions models.
- Integration3d ago· GitHub Releases
langchain-anthropic==1.7.5
LangChain Anthropic 1.7.5 adds Claude Sonnet 5.5 compatibility. It also fixes replay serialization so invalid tool calls are represented as tool use.
- Integration3d ago· GitHub Releases
langchain-core==1.6.6
LangChain Core 1.6.6 adds Claude Sonnet 5.5 compatibility and corrects docstring examples that previously failed when copied.
- Feature4d ago· GitHub Releases
langchain-fireworks==1.7.0
LangChain Fireworks 1.7.0 adds prompt caching middleware. It also improves classification of mid-stream read timeouts and refreshes model profile data.
- Integration4d ago· GitHub Releases
langchain==1.4.3
LangChain 1.4.3 adds Bedrock Mantle chat model support and recognizes GPT-6 structured output without profiles. It also fixes invalid agent tool calls and sanitizes cache settings for fallback models.
- Feature7d ago· GitHub Releases
langchain-fireworks==1.6.3
LangChain Fireworks 1.6.3 marks native PDF inputs as unsupported, preserves malformed tool arguments as diagnostic JSON, and refreshes model profile data.
- Feature8d ago· GitHub Releases
langchain-core==1.6.5
LangChain Core 1.6.5 fixes XML buffer string handling by abbreviating long tool IDs. This is a maintenance release following version 1.6.4.
- Feature8d ago· GitHub Releases
langchain-openai==1.6.6
langchain-openai 1.6.6 now raises errors received through the streaming event path instead of potentially leaving them unreported.
- Feature9d ago· GitHub Releases
langchain-anthropic==1.7.4
langchain-anthropic 1.7.4 adds support for changing tools mid-conversation through SystemMessage. It also updates model profiles and fixes an integration test cassette.
- Feature9d ago· GitHub Releases
langchain-openai==1.6.5
langchain-openai 1.6.5 adds support for changing tools mid-conversation through SystemMessage. It also adds profile augmentations for Opus 5.5 and GPT-6.
- Feature10d ago· GitHub Releases
langchain-openai==1.6.4
LangChain released langchain-openai 1.6.4. The update refreshes OpenAI model profile data.
- Integration10d ago· GitHub Releases
langchain-fireworks==1.6.2
LangChain released langchain-fireworks 1.6.2, updating model references used in LLM tests, refreshing model profiles, and bumping several dependencies.
- Integration10d ago· GitHub Releases
langchain-deepseek==1.1.1
LangChain released langchain-deepseek 1.1.1 with compatible minimum OpenAI dependencies, strict-mode routing through DeepSeek’s beta endpoint, and corrected prompt-cache token mapping.
- Integration10d ago· GitHub Releases
langchain-openrouter==0.2.9
LangChain released langchain-openrouter 0.2.9 with refreshed model profile data and an AnyIO dependency update to 4.14.2.
- Feature11d ago· GitHub Releases
langchain-openai==1.6.3
LangChain OpenAI 1.6.3 exposes inferred Responses API routing during initialization and adds support for GPT-6 request constraints. It also updates the anyio dependency.
- Deprecation11d ago· GitHub Releases
langchain-core==1.6.4
LangChain Core 1.6.4 deprecates chat message history and updates its anyio and soupsieve dependencies.
- Feature12d ago· GitHub Releases
langchain-typesafe==0.0.1a3
LangChain released the initial alpha of langchain-typesafe, adding TypeSafeClassifier and experimental AutoModeMiddleware and ModelRouterMiddleware. The release also makes classifier questions invocation-scoped and fixes tracing of usage metadata.
- Feature14d ago· GitHub Releases
langchain==1.4.2
LangChain 1.4.2 fixes human-in-the-loop tool call editing so model-generated tool calls are preserved. It also adds a notice to ToolMessage.
- News15d ago· GitHub Releases
langchain-typesafe==0.0.1a2
Initial release release(typesafe): bump to 0.0.1a2 (#40576) feat(typesafe): experimental `AutoModeMiddleware` (#40545) feat(typesafe): experimental `ModelRouterMiddleware` (#40543) fix(typesafe): trace usage metadata (#40570) feat(typesafe): `TypeSafeClassifier` (#40542)
- Feature15d ago· GitHub Releases
langchain-typesafe==0.0.1a1
LangChain released langchain-typesafe 0.0.1a1, the package’s initial alpha version. It introduces TypeSafeClassifier.
- Feature16d ago· GitHub Releases
langchain==1.4.1
LangChain 1.4.1 fixes preservation of open MCP object arguments and corrects the InterruptOnConfig documentation. It is a patch release following version 1.4.0.
- Feature21d ago· GitHub Releases
langchain-core==1.6.3
LangChain Core 1.6.3 lets gateways override model name and provider metadata used for tracing. The release also adds deprecated-path test coverage and cleans up stale API documentation.
- Feature22d ago· GitHub Releases
langchain-anthropic==1.7.2
LangChain Anthropic 1.7.2 fixes handling of invalid tool-use blocks so they are preserved rather than discarded or altered.
- Feature23d ago· GitHub Releases
langchain-openai==1.6.2
LangChain OpenAI 1.6.2 adds reasoning-effort support for GPT-6 Astra and updates its httpx2 dependency from 2.10.0 to 2.12.0.
- Integration24d ago· GitHub Releases
langchain-openai==1.6.1
langchain-openai 1.6.1 adds async tool and configuration_update support, Azure AD authentication compatibility with OpenAI 3.8, and Responses API routing for gpt-5.6-sol. It also corrects reasoning profiles for GPT-5 models.
- Feature28d ago· GitHub Releases
langchain-core==1.6.2
LangChain Core 1.6.2 adds asynchronous tool support for OpenAI and prevents content mutation in Google GenAI and Amazon Bedrock Converse integrations. It also updates Mistune and Tornado dependencies.
- Integration29d ago· GitHub Releases
langchain==1.4.0
LangChain 1.4.0 adds a langchain.mcp namespace and MCPAdapter, with runnable MCP examples. It also fixes agent tool routing to include the model destination and reduces Anthropic middleware trace inputs.
- Feature29d ago· GitHub Releases
langchain-anthropic==1.7.1
langchain-anthropic 1.7.1 adds support for Claude Fable 5.1 and omits middleware trace inputs to improve performance.
- FeatureSep 2· GitHub Releases
langchain==1.4.0a4
LangChain 1.4.0a4 updates its MCP adapter for FastMCP 4.x, including protocol-aware elicitation and interrupt routing. The alpha also revises MCP tool APIs, metadata, caching controls, and client-group handling.
- FeatureSep 1· GitHub Releases
langchain==1.4.0a3
LangChain 1.4.0a3 introduces the langchain.mcp namespace for adapting FastMCP servers and clients into LangChain tools. It adds tool discovery caching, MCP metadata, single-tool conversion, and LangGraph interrupts for server elicitation requests.
- FeatureAug 28· GitHub Releases
langchain==1.4.0a2
LangChain 1.4.0a2 adds an alpha langchain.mcp adapter that converts MCP servers into tools for create_agent. Built on FastMCP, it supports URLs, local scripts, in-process servers, multi-server configurations, and custom clients through one interface.
- FeatureAug 27· GitHub Releases
langchain==1.4.0a1
LangChain 1.4.0a1 is an initial alpha release focused on MCP adapter behavior, including elicitation types, protocol compatibility across servers, and clearer failures for private FastMCP dependencies.
- IntegrationAug 27· GitHub Releases
langchain-fireworks==1.6.1
LangChain Fireworks 1.6.1 removes reasoning-history blocks and refreshes model profile data for the Fireworks integration.
- NewsAug 27· GitHub Releases
langchain==1.3.18
LangChain 1.3.18 preserves content-block structure during PII middleware redaction and improves indexing for GenAI v1 streaming content.
- FeatureAug 25· GitHub Releases
langchain==1.3.17
LangChain 1.3.17 fixes handling of custom rejection reasons in human-in-the-loop workflows and updates dependencies.
- FeatureAug 21· GitHub Releases
langchain-perplexity==1.4.1
LangChain’s Perplexity package 1.4.1 fixes Responses API message typing, preserves caller-supplied extra_body parameters, and switches to a supported model. The broader release also adds package-version tracing metadata and streaming tool-call validation.
- IntegrationAug 20· GitHub Releases
langchain-fireworks==1.6.0
LangChain’s Fireworks integration 1.6.0 adds document reranking and standard model exceptions. It also filters invalid tool calls, prevents duplicate service-tier parameters, and refreshes model profiles and dependencies.
- FeatureAug 20· GitHub Releases
langchain==1.3.16
LangChain 1.3.16 adds standard model exception types and custom token counters for ContextEditingMiddleware. It also corrects retry handling and repeated schema ordering.
- IntegrationAug 20· GitHub Releases
langchain-anthropic==1.6.1
LangChain’s Anthropic integration 1.6.1 filters invalid tool calls from v1 content. This patch should reduce failures when processing malformed Anthropic tool-call responses.
- FeatureAug 19· GitHub Releases
langchain-openai==1.6.0
LangChain OpenAI 1.6.0 introduces standardized model exception types and clearer errors for unexpected chat response types. The changes should simplify consistent error handling in applications using LangChain model integrations.