On behalf of the Spring AI engineering team and everyone who has contributed, I'm happy to announce that Spring AI 1.1.0-M3 has been released and is now available from Maven Central.

This milestone release focuses primarily on Model Context Protocol (MCP) enhancements, incorporating the MCP Java SDK v0.14.0 upgrade along with new resource template capabilities and security documentation.

Release Summary

This release includes 46 improvements, bug fixes, and documentation updates. Key focus areas include:

• Model Context Protocol: Upgraded to MCP Java SDK 0.14.0 with resource template support and security best practices
• New Features: 7 capabilities including Azure Cosmos DB chat memory and GemFire metadata filtering
• Stability: 11 bug fixes addressing community-reported issues
• Documentation: 9 improvements including MCP security guidance
• Dependency Upgrades: 3 updates including security fixes (CVE-2025-54988)

🔧 Model Context Protocol Improvements

This release brings enhanced MCP integration to Spring AI, motivated by significant improvements in the MCP Java SDK v0.14.0:

Core MCP Enhancements

• Updated MCP Java SDK: Enhanced resource template management and API consistency
• Security Documentation: Added MCP security reference documentation for secure deployment configurations
• Resource Template Support: Added parameterized resource template capabilities for both sync and async MCP servers, enabling flexible resource provisioning with dynamic parameters
• Client-Side Validation: New tool output schema validation and caching capabilities for improved reliability
• Robust Error Handling: Better resilience for MCP server interactions with proper handling of non-compliant notification responses and Content-Length: 0 scenarios
• Spec Compliance: Proper resource not found handling according to MCP specification
• API Refinements: Improved JSON type handling, consistent naming conventions (MCP_SESSION_ID), and optional lastModified field support in Annotations

For developers using MCP in Spring AI applications, this release provides a more stable and feature-rich foundation for tool integration workflows.

Special thanks to the MCP Java SDK community for their exceptional work on the underlying SDK improvements that made this Spring AI release possible:

Broadcom: Christian Tzolov (@tzolov), Daniel Garnier-Moiroux (@Kehrlann), Dariusz Jędrzejczyk (@chemicL)
Oracle: Sergio del Amo (@sdelamo)
Confluent: Pascal Vantrepote (@pascalconfluent)
Amazon Web Services: Anurag Pant (@pantanurag555)
Open Source Community: Liujunjie (@JunJieLiu51520) - First-time contributor to MCP Java SDK

Additional Functional Areas Enhanced

Beyond MCP improvements, this release brings enhancements across major functional areas of Spring AI:

• Azure Cosmos DB Chat Memory - Added Azure Cosmos DB integration for chat memory storage, expanding beyond existing Cassandra support
• Anthropic Prompt Caching - Updated Anthropic integration with prompt caching strategies (system-only, system-and-tools, conversation-history) and support for the latest Claude models (Sonnet 4.5, Opus 4.1) with consistent naming conventions
• GemFire Vector Search - Added metadata filtering for GemFireVectorStore enabling similarity search queries with filtering conditions for RAG applications
• MarkdownDocumentReader - Now processes multiple documents in a single operation for batch document ingestion
• Mistral AI Improvements - Builder pattern support across the Mistral module plus improved JsonSchemaGenerator handling for function calling parameters

Looking Ahead: Spring AI 1.1 GA

As we progress toward the Spring AI 1.1 General Availability release, the team is focused on three key areas:

Model Context Protocol (MCP) - Continued MCP enhancements

Chat Model Features - Expanding prompt caching and thinking/reasoning mode support across model providers

Chat Memory - Addressing community-reported chat memory issues

For the latest updates and to contribute to the project, visit our GitHub repository or join the discussion in our community channels.

Resources

Project Page | GitHub | Issues | Documentation | Stack Overflow

🙏 Contributors

Thanks to all contributors who made this release possible:

• ashakirin (@andrei.shakirin)
• Cameron Kirk (@kirkster96)
• cho-thinkfree-com (@cho)
• Christian Tzolov (@tzolov)
• codeboyzhou (@imzhouchen)
• Daniel Garnier-Moiroux (@git)
• David Frizelle (@dafriz)
• Dmitrii Chechetkin (@dmitrii.chechetkin)
• Eric Bottard (@bottarde)
• Hyoseop Song (@crad_on25)
• Ilayaperumal Gopinathan (@ilayaperumalg)
• James Ward (@james)
• Jason Huynh (@jason.huynh)
• Jason Smith (@jasonparallel)
• lance (@leehaut)
• Mark Pollack (@markpollack)
• NathanGrand (@nathangrand)
• Nils Breunese (@nils)
• Oleksandr Klymenko (@alexanderklmn)
• Soby Chacko (@soby.chacko)
• Sun Yuhan (@sunyuhan1998)
• Theo van Kraay (@theo.van)
• Tran Ngoc Nhan (@ngocnhan.tran1996)
• Yanming Zhou (@zhouyanming)
• YuJie Wan (@eeaters)
• YunKui Lu (@luyunkui95)