LiteLLM

This release addresses several critical bug fixes across various integrations and internal components, including memory leak resolutions and stability improvements for Python versions 3.9-3.12.3. It also resolves issues with span export, trace data handling, and LLM observability.

This release introduces significant feature enhancements across MLFlow integration, AI Guard, LLM Observability, and adds process tag propagation across many components. Several bug fixes address issues in CI Visibility, tracing context handling, and profiler stability.

This release introduces a new REST endpoint for prompt metadata updates and an expression filter DSL for sessions. It also includes bug fixes for cost tracking and UI rendering.

This release introduces REST CRUD operations for API keys, enhances the CLI setup process, and adds a user friction evaluator. It also includes various UI and bug fixes for a more robust user experience.

Version 2.12.0 introduces support for structured data inputs in evals and patches several bugs, including a security-related pin on the LiteLLM dependency.

This patch release addresses a security concern by pinning the LiteLLM dependency to a safe version.

This release addresses critical bugs related to context propagation in LangChain/LangGraph and fixes issues with recording OpenAI chat completions, especially when streaming via LiteLLM.

DSPy 3.2.1 focuses on bug fixes, particularly around async streaming and embedding calls, and includes numerous documentation and CI/CD improvements. The upper bound constraint on the litellm dependency has been removed.

This release introduces async tool spec support across several integrations, adds new LLM providers like AI Badgr and Typecast, and provides significant updates to MongoDB and Vertex AI vector stores. It also includes various bug fixes for metadata filtering and Pydantic schema handling.

This release introduces YugabyteDB chat storage support and Bedrock prompt caching, alongside critical bug fixes for path handling, LiteLLM context calculations, and Milvus node creation.

This release introduces several new integrations including ElevenLabs and ApertureDB, adds support for OpenAI JSON Schema and Ollama's 'think' feature, and includes a breaking rename of the JsonPickleSerializer to PickleSerializer.

This release introduces a major memory system revamp, adds several new integrations including Gel, Oxylabs, and Meta Llama-api, and provides critical bug fixes for ReAct agents and tool call parsing across multiple LLM providers.

This release introduces a new CLI config command group, initial tracing capabilities, and LangDB integration while removing the deprecated User Memory system. It also includes performance optimizations and improved CLI error reporting.

This release introduces Bedrock agent toolkits and the SerperScrapeWebsiteTool while upgrading Mem0 Storage to v2. It also includes critical fixes for Chroma initialization, Ollama message handling, and removes legacy SQLite workarounds.

This release introduces LiteAgent with Guardrail integration, async tool execution, and support for Python 3.13. It also includes several core fixes for usage metrics, telemetry, and CLI multi-org actions.

This release introduces Managed Identity for Azure AI Search and OAuth2 for Jira, while deprecating SingleStore and Azure Cosmos DB NoSQL integrations in favor of dedicated partner packages. It also includes critical cost calculation fixes for OpenAI o3/o4 models and removes the pandas dependency for DuckDB vector searches.