v2.27.0

Deepset has launched Haystack v2.27.0, a significant update to its open-source framework for building production-ready AI applications like RAG (retrieval-augmented generation) systems and AI agents. The headline feature is 'implicit list joining' in pipelines, which automatically combines multiple inputs into a single list for a component. This eliminates the need for extra components to manually join data, simplifying complex workflows. For example, developers can now directly combine a plain query string with a list of ChatMessage objects into a single list[ChatMessage] input, streamlining the construction of conversational AI pipelines.

The update also brings major quality-of-life improvements for developers. The InMemoryDocumentStore now includes metadata inspection and filtering utilities previously only available in other document stores, allowing for seamless local prototyping and debugging. Furthermore, the AzureOpenAIChatGenerator component now exposes a `SUPPORTED_MODELS` variable, officially listing newer models like `gpt-5-mini` and `gpt-4o`. Other notable additions include partial multimodal support for models like Qwen 3.5 in the HuggingFaceLocalChatGenerator, new async filter helpers, and a security fix in the ChatPromptBuilder to prevent template variables from being misinterpreted as structured content.