v4.5.1
mudler/LocalAIv4.5.1Jun 26, 2026by mudler
AI Summary
A comprehensive release with extensive macOS/Metal support, UI enhancements, and backend updates across multiple components.
Key Highlights
- macOS/Metal support for purego Go backends and multiple C++ backends
- UI usage & UX enhancements including last-used model, polling, and accessibility
- Darwin/Metal builds for vision C++ backends (depth-anything, locate-anything, rfdetr-cpp, sam3-cpp)
- Single-build ggml CPU_ALL_VARIANTS for llama-cpp + turboquant (x86/arm64/apple)
- Darwin/Metal build for trl backend
New Features
- macOS/Metal support for supertonic backend
- macOS/Metal build for liquid-audio backend
- macOS/Metal support via vllm-metal (MLX)
- CPU_MOE/n_CPU_MEO options for llama-cpp
- Darwin/Metal builds for vision C++/ggml backends
- Darwin build for localvqe backend (acoustic echo cancellation)
- Darwin/Metal build for privacy-filter backend
Full Release Notes
<!-- Release notes generated using configuration in .github/release.yml at master --> ## What's Changed ### Other Changes * chore(model gallery): :robot: add 1 new models via gallery agent by @localai-bot in https://github.com/mudler/LocalAI/pull/10472 * chore: :arrow_up: Update ikawrakow/ik_llama.cpp to `7ccf1d209588962b96eacca325b37e9b3e8faf5e` by @localai-bot in https://github.com/mudler/LocalAI/pull/10456 * chore: :arrow_up: Update CrispStrobe/CrispASR to `96b2a6ee31d30389fed8a7ef1a54239b75231ddc` by @localai-bot in https://github.com/mudler/LocalAI/pull/10465 * chore: :arrow_up: Update ggml-org/llama.cpp to `be4a6a63eb2b848e19c277bdcf2bd399e8af76d9` by @localai-bot in https://github.com/mudler/LocalAI/pull/10467 * chore: :arrow_up: Update ggml-org/whisper.cpp to `43d78af5be58f41d6ffbc227d608f104577741ea` by @localai-bot in https://github.com/mudler/LocalAI/pull/10466 * chore: :arrow_up: Update mudler/parakeet.cpp to `89f5e2977b4d8bccd45e7bcc6f2ef7c4ed49e89a` by @localai-bot in https://github.com/mudler/LocalAI/pull/10468 * fix(agents): URL-decode collection/agent name path params (#10443) by @localai-bot in https://github.com/mudler/LocalAI/pull/10471 * fix(distributed): track in-flight for SoundDetection requests by @localai-bot in https://github.com/mudler/LocalAI/pull/10475 * refactor(distributed): make in-flight tracking coverage a compile-time contract by @localai-bot in https://github.com/mudler/LocalAI/pull/10476 * fix(pii): load default detectors at startup + add LOCALAI_PII_DEFAULT_DETECTORS by @richiejp in https://github.com/mudler/LocalAI/pull/10474 * i18n(id): update and complete Indonesian translations by @dedyf5 in https://github.com/mudler/LocalAI/pull/10480 * fix(realtime): resolve model aliases for pipeline sub-models by @localai-bot in https://github.com/mudler/LocalAI/pull/10484 * fix(backends): darwin/metal support for supertonic by @localai-bot in https://github.com/mudler/LocalAI/pull/10488 * feat(backends): add darwin/metal build for liquid-audio by @localai-bot in https://github.com/mudler/LocalAI/pull/10486 * chore(model-gallery): :arrow_up: update checksum by @localai-bot in https://github.com/mudler/LocalAI/pull/10495 * docs: :arrow_up: update docs version mudler/LocalAI by @localai-bot in https://github.com/mudler/LocalAI/pull/10491 * feat(ui): usage & UX enhancements (last-used model, polling, starter models, usage cost, a11y) by @localai-bot in https://github.com/mudler/LocalAI/pull/10496 * fix(config): per-device VRAM headroom for Blackwell defaults (#10485) by @localai-bot in https://github.com/mudler/LocalAI/pull/10494 * feat(ui): data-driven hardware model recommendations + gallery surfacing by @localai-bot in https://github.com/mudler/LocalAI/pull/10500 * chore: :arrow_up: Update ikawrakow/ik_llama.cpp to `d5507e33ae7ee2b7b41475f08044d3bde3b839ee` by @localai-bot in https://github.com/mudler/LocalAI/pull/10498 * chore: :arrow_up: Update ServeurpersoCom/omnivoice.cpp to `0f37401bebe9b20c0160a888e592108fc1d17607` by @localai-bot in https://github.com/mudler/LocalAI/pull/10492 * fix(backends): darwin/metal support across purego Go backends by @localai-bot in https://github.com/mudler/LocalAI/pull/10481 * feat(backends): add darwin/metal (MPS) build for trl by @localai-bot in https://github.com/mudler/LocalAI/pull/10487 * feat(llama-cpp): cpu_moe/n_cpu_moe options + generic upstream-flag passthrough by @localai-bot in https://github.com/mudler/LocalAI/pull/10490 * chore: :arrow_up: Update ServeurpersoCom/qwentts.cpp to `9dbe7ea26a01b30fccb117ae5e86807c1dc23d42` by @localai-bot in https://github.com/mudler/LocalAI/pull/10499 * fix: correct scheme/host on self-referential URLs behind an HTTPS reverse proxy (#10482) by @localai-bot in https://github.com/mudler/LocalAI/pull/10504 * chore: :arrow_up: Update ggml-org/llama.cpp to `8be759e6f70d629638a7eb70db3824cbdcea370b` by @localai-bot in https://github.com/mudler/LocalAI/pull/10501 * chore: :arrow_up: Update leejet/stable-diffusion.cpp to `8caa3f908ae6d4a4bef531e73b9a969f266a3d1f` by @localai-bot in https://github.com/mudler/LocalAI/pull/10493 * chore(model gallery): :robot: add 1 new models via gallery agent by @localai-bot in https://github.com/mudler/LocalAI/pull/10505 * feat(vllm): macOS/Metal support via vllm-metal (MLX) by @localai-bot in https://github.com/mudler/LocalAI/pull/10489 * feat: single-build ggml CPU_ALL_VARIANTS for llama-cpp + turboquant (x86/arm64/apple) by @localai-bot in https://github.com/mudler/LocalAI/pull/10497 * fix(config): gate parallel-slot default on per-device VRAM too (#10485) by @localai-bot in https://github.com/mudler/LocalAI/pull/10507 * fix(auth): make advisory locks dialect-aware and harden SQLite DSN by @localai-bot in https://github.com/mudler/LocalAI/pull/10509 * feat(backends): darwin/Metal builds for vision C++/ggml backends (depth-anything, locate-anything, rfdetr-cpp, sam3-cpp) by @localai-bot in https://github.com/mudler/LocalAI/pull/10511 * feat(backends): darwin build for the localvqe backend (acoustic echo cancellation) by @localai-bot in https://github.com/mudler/LocalAI/pull/10512 * docs(backends): make OS coverage explicit + require darwin support for new backends by @localai-bot in https://github.com/mudler/LocalAI/pull/10516 * chore: :arrow_up: Update ikawrakow/ik_llama.cpp to `b84902d2ad27c34f989f23947200c4b91b1568fd` by @localai-bot in https://github.com/mudler/LocalAI/pull/10515 * chore: bump localrecall for postgres per-connection timeouts by @localai-bot in https://github.com/mudler/LocalAI/pull/10517 * chore: pin localrecall to tagged v0.6.3 by @localai-bot in https://github.com/mudler/LocalAI/pull/10518 * fix(backends): quote $CURDIR in run.sh (fixes backends in paths with spaces) by @localai-bot in https://github.com/mudler/LocalAI/pull/10519 * chore: :arrow_up: Update CrispStrobe/CrispASR to `8f1218141b792b8868861c1af17ba1e361b05dc0` by @localai-bot in https://github.com/mudler/LocalAI/pull/10502 * chore: :arrow_up: Update ggml-org/llama.cpp to `9d5d882d8cd0f0a9283d87ed5e6fe3ee0d925fb1` by @localai-bot in https://github.com/mudler/LocalAI/pull/10514 * feat(backends): darwin/Metal build for the privacy-filter backend by @localai-bot in https://github.com/mudler/LocalAI/pull/10513 * feat(backends): make PreferDevelopmentBackends install the development image as primary by @localai-bot in https://github.com/mudler/LocalAI/pull/10520 * chore(model gallery): :robot: add 1 new models via gallery agent by @localai-bot in https://github.com/mudler/LocalAI/pull/10521 * fix(backends): ship the package/ dir for darwin go backend images by @localai-bot in https://github.com/mudler/LocalAI/pull/10522 **Full Changelog**: https://github.com/mudler/LocalAI/compare/v4.5.0...v4.5.1