LocalAI Releases
155 releases of mudler/LocalAI
- v4.5.5
Bug fix release addressing CI build issues and adding a new model.
Jun 27, 2026
- v2026.626.0
This release integrates Hermes as a first-class adapter and introduces task watchdogs and 'ask' work mode. It improves workspace management, sandbox visibility, and agent configuration.
Jun 27, 2026
- v4.5.4
A minor bug fix release focusing on macOS/Darwin binary derivation for the backend runner.
Jun 27, 2026
- v4.5.3
Focuses on macOS packaging and backend fixes, including binary signing, rpath configuration, and GPU library bundling.
Jun 26, 2026
- v4.5.2
Fixes macOS/Darwin opus backend build and packaging to ensure it can be built and packaged on macOS systems.
Jun 26, 2026
- v4.5.1
A comprehensive release with extensive macOS/Metal support, UI enhancements, and backend updates across multiple components.
Jun 26, 2026
- v4.5.0
This release introduces new backends for depth perception, sound classification, and on-device TTS. It features a redesigned UI, enhanced real-time voice capabilities, and enables multi-user serving by default with optimized continuous batching.
Jun 23, 2026
- v2026.618.0
Introduces a Skills Store, self-hostable sandbox execution, and major security improvements for multi-tenant isolation.
Jun 18, 2026
- v0.7.0
A major release (242 PRs) introducing a new web console, production-grade OpenAI-compatible Batch API, and first-class support for TensorRT-LLM.
Jun 18, 2026
- v0.103.5
This release focuses on UI/UX improvements for the test set annotation queue and playground, alongside fixes for Redis queue behavior and documentation updates.
Jun 15, 2026
- v4.4.3
Adds new backends including object detection, Indonesian language support, and multiple model updates.
Jun 13, 2026
- v0.103.4
Includes new SMTP support, cleanup of git-based routes, and critical fixes for broken workflow, revision, and version data.
Jun 13, 2026
- v0.103.3
Focuses on frontend playground stability, evaluator improvements, and new account management features like bulk archiving and self-service deletion.
Jun 12, 2026
- v4.4.2
Focuses on bug fixes for vLLM compatibility and CUDA development package installation.
Jun 11, 2026
- v4.4.1
Restores compatibility with vLLM >= 0.22 and adds streaming capabilities to the realtime pipeline.
Jun 11, 2026
- lingo.dev@0.136.4
This patch release improves the parsing and error handling for Bring Your Own Key (BYOK) models within the `run` and `i18n` commands to improve reliability.
Jun 11, 2026
- v0.103.2
This release introduces several frontend improvements including sticky headers in JSON views, multiple metric support for evaluators, and a decoupled playground chat behavior.
Jun 11, 2026
- v4.4.0
A major multimodal-and-distributed release featuring two new ASR backends, intelligent middleware, distributed mode improvements, video capabilities, and object detection.
Jun 10, 2026
- v0.103.1
This update focuses on frontend refinements, including improved deployment date display, a redesigned configuration tab for evaluation runs, and new entity reference chips with hovercards.
Jun 10, 2026
- @lingo.dev/_compiler@0.12.5
This is a dependency update for the compiler package that bumps the SDK version to 0.16.5 to ensure compatibility and stability.
Jun 9, 2026