The Rspamd team is pleased to announce the release of Rspamd 3.11.1. An overview of the changes in this release is presented below:

GPT plugin improvements

• Added support for setting reasons for classification
• Added compatibility with gpt4all, ollama
• Added support for multiple model consensus

RBL module fixes

• Removed nixspam RBL
• Added support for disabling RBLs via map
• Fixed content_urls and images checks
• Fixed ipv4/ipv6 settings when used with resolve_ip
• Allow images to be set inside checks

Other features

• New redis caching framework for Lua plugins
• Added ability to create timers via Lua
• Support hashing arbitrary Lua types
• Support LLM-based anonymization in rspamadm mime command
• Per-key read_only support for fuzzy
• Allow multiple Lua scripts for fuzzy storage
• Support UTF8 sub and len variants for rspamd_text

Other fixes

• Fixed adding footer to HTML parts with inline images
• Better support for inline maps
• Ham and spam sets for neural module are now expired
• Fix crash when Rspamd is built without Hyperscan

See here for the full list of changes. Thank you to everyone who contributed to this release.