Memory

This release includes numerous bug fixes across various llama-index components, improving stability and correctness. It also introduces new features such as support for new LLM models and enhanced multimodal capabilities.

This release primarily focuses on dependency updates across numerous packages via mass uv lock upgrades and introduces Multimodal synthesis support in llama-index-core. Several minor bug fixes address issues related to instrumentation, caching, and token counting.

This release introduces distributed data ingestion via Ray, token-based code splitting, and new integrations like Apertis LLM and Volcengine MySQL vector store. Notably, ChatMemoryBuffer has been replaced by the generic Memory class.

This release introduces Baseten and ZenRows integrations, adds MMR search to Chroma, and updates IBM integrations with external URL support while deprecating instance_id. It also includes critical fixes for ContextChatEngine and Postgres vector store deletion logic.

This release introduces the Vectorize retriever and DeSearch tool, while providing critical bug fixes for tool call parsing in Bedrock and OpenAI, and improving multi-agent workflow stability.

This release introduces significant features including model client caching, GraphRAG integration, Semantic Kernel adapters, and a new agent memory interface. It also expands declarative configuration support and brings back the Jupyter code executor.