v3.9.0

[mudler]

Xmas-release 🎅 LocalAI 3.9.0! 🚀

LocalAI 3.9.0 is focused on stability, resource efficiency, and smarter agent workflows. We've addressed critical issues with model loading, improved system resource management, and introduced a new Agent Jobs panel for scheduling and managing background agentic tasks. Whether you're running models locally or orchestrating complex agent workflows, this release makes it faster, more reliable, and easier to manage.

📌 TL;DR

Feature	Summary
Agent Jobs Panel	Schedule and run background tasks with cron or via API — perfect for automated workflows.
Smart Memory Reclaimer	Automatically frees up GPU/VRAM by evicting least recently used models when memory is low.
LRU Model Eviction	Models are automatically unloaded from memory based on usage patterns to prevent crashes.
MLX & CUDA 13 Support	New model backends and enhanced GPU compatibility for modern hardware.
UI Polish & Fixes	Cleaned-up navigation, fixed layout overflow, and various improvements.
Vibevoice	Added support for the vibevoice backend!

---

🚀 New Features

🤖 Agent Jobs Panel: Schedule & Automate Tasks

LocalAI 3.9.0 introduces a new Agent Jobs panel in the web UI and API, allowing you to create, run, and schedule agentic tasks in the background that can be started programmatically via API or from the Web interface.

• Run agent prompts on a schedule using cron syntax, or via API.
• Agents are defined via the model settings, supporting MCP.
• Trigger jobs via API for integration into CI/CD or external tools.
• Optionally send results to a webhook for post-processing.
• Templates and prompts can be dynamically populated with variables.

✅ Use cases: Daily reports, CI integration, automated data processing, scheduled model evaluations.

---

🧠 Smart Memory Reclaimer: Auto-Optimize GPU Resources

We’ve introduced a new Memory Reclaimer that monitors system memory usage and automatically frees up GPU/VRAM when needed.

• Tracks memory consumption across all backends.
• When usage exceeds a configured threshold, it evicts the least recently used (LRU) models.
• Prevents out-of-memory crashes and keeps your system stable during high load.

This is a foundational step toward adaptive resource management — future versions will expand this with more advanced policies and fleet-wide control.

---

🔁 LRU Model Eviction: Intelligent Model Management

Building on the new reclaimer, LocalAI now supports LRU (Least Recently Used) eviction for loaded models.

• Set a maximum number of models to keep in memory (e.g., limit to 3).
• When a new model is loaded and the limit is reached, the oldest unused model is automatically unloaded.
• Fully compatible with single_active_backend mode (now defaults to LRU=1 for backward compatibility).

💡 Ideal for servers with limited VRAM or when running multiple models in parallel.

---

🖥️ UI & UX Polish

• Fixed navbar ordering and login icon — clearer navigation and better visual flow.
• Prevented tool call overflow in chat view — no more clipped or misaligned content.
• Uniformed link paths (e.g., /browse/ instead of browse) for consistency.
• Fixed model selection toggle — header updates correctly when switching models.
• Consistent button styling — uniform colors, hover effects, and accessibility.

---

📦 Backward Compatibility & Architecture

• Dropped x86_64 Mac support: no longer maintained in GitHub Actions; ARM64 (M1/M2/M3/M4) is now the recommended architecture.
• Updated data storage path from /usr/share to /var/lib: follows Linux conventions for mutable data.
• Added CUDA 13 support: now available in Docker images and L4T builds.
• New VibeVoice TTS backend real-time text-to-speech with voice cloning support. You can install it from the model gallery!
• StableDiffusion-GGML now supports LoRA: expand your image-generation capabilities.

---

🛠️ Fixes & Improvements

• Issue: After v3.8.0, /readyz and /healthz endpoints required authentication, breaking Docker health checks and monitoring tools
• Issue: Fixed crashes when importing models from Hugging Face URLs with subfolders (e.g., huggingface://user/model/GGUF/model.gguf).

---

🚀 The Complete Local Stack for Privacy-First AI

LocalAI	The free, Open Source OpenAI alternative. Drop-in replacement REST API compatible with OpenAI specifications for local AI inferencing. No GPU required. Link: https://github.com/mudler/LocalAI
LocalAGI	Local AI agent management platform. Drop-in replacement for OpenAI's Responses API, supercharged with advanced agentic capabilities and a no-code UI. Link: https://github.com/mudler/LocalAGI
LocalRecall	RESTful API and knowledge base management system providing persistent memory and storage capabilities for AI agents. Works alongside LocalAI and LocalAGI. Link: https://github.com/mudler/LocalRecall

---

❤️ Thank You

LocalAI is a true FOSS movement — built by contributors, powered by community.

If you believe in privacy-first AI:

• ✅ Star the repo
• 💬 Contribute code, docs, or feedback
• 📣 Share with others

Your support keeps this stack alive.

---

✅ Full Changelog

📋 Click to expand full changelog

What's Changed

Breaking Changes 🛠

• chore: switch from /usr/share to /var/lib for data storage by @poretsky in chore: switch from /usr/share to /var/lib for data storage #7361
• chore: drop drawin-x86_64 support by @mudler in chore: drop drawin-x86_64 support #7616

Bug fixes 🐛

• fix: do not require auth for readyz/healthz endpoints by @mudler in fix: do not require auth for readyz/healthz endpoints #7403
• fix(ui): navbar ordering and login icon by @mudler in fix(ui): navbar ordering and login icon #7407
• fix: configure sbsa packages for arm64 by @mudler in fix: configure sbsa packages for arm64 #7413
• fix(ui): prevent box overflow in chat view by @mudler in fix(ui): prevent box overflow in chat view #7430
• fix(ui): Update few links in web UI from 'browse' to '/browse/' by @rampa3 in fix(ui): Update few links in web UI from 'browse' to '/browse/' #7445
• fix(paths): remove trailing slash from requests by @mudler in fix(paths): remove trailing slash from requests #7451
• fix(downloader): do not download model files if not necessary by @mudler in fix(downloader): do not download model files if not necessary #7492
• fix(config): make syncKnownUsecasesFromString idempotent by @mudler in fix(config): make syncKnownUsecasesFromString idempotent #7493
• fix: make sure to close on errors by @mudler in fix: make sure to close on errors #7521
• fix(llama.cpp): handle corner cases with tool array content by @mudler in fix(llama.cpp): handle corner cases with tool array content #7528
• fix(7355): Update llama-cpp grpc for v3 interface by @sredman in fix(7355): Update llama-cpp grpc for v3 interface #7566
• fix(chat-ui): model selection toggle and new chat by @mudler in fix(chat-ui): model selection toggle and new chat #7574
• fix: improve ram estimation by @mudler in fix: improve ram estimation #7603
• fix(ram): do not read from cgroup by @mudler in fix(ram): do not read from cgroup #7606
• fix: correctly propagate error during model load by @mudler in fix: correctly propagate error during model load #7610
• fix(ci): remove specific version for grpcio packages by @mudler in fix(ci): remove specific version for grpcio packages #7627
• fix(uri): consider subfolders when expanding huggingface URLs by @mintyleaf in fix(uri): consider subfolders when expanding huggingface URLs #7634

Exciting New Features 🎉

• feat: agent jobs panel by @mudler in feat: agent jobs panel #7390
• chore: refactor css, restyle to be slightly minimalistic by @mudler in chore: refactor css, restyle to be slightly minimalistic #7397
• feat(hf-api): return files in nested directories by @mudler in feat(hf-api): return files in nested directories #7396
• feat(agent-jobs): add multimedia support by @mudler in feat(agent-jobs): add multimedia support #7398
• feat: add cuda13 images by @mudler in feat: add cuda13 images #7404
• fix: use ubuntu 24.04 for cuda13 l4t images by @mudler in fix: use ubuntu 24.04 for cuda13 l4t images #7418
• feat(diffusers): implement dynamic pipeline loader to remove per-pipeline conditionals by @Copilot in feat(diffusers): implement dynamic pipeline loader to remove per-pipeline conditionals #7365
• chore(importers/llama.cpp): add models to 'llama-cpp' subfolder by @mudler in chore(importers/llama.cpp): add models to 'llama-cpp' subfolder #7450
• feat(vibevoice): add new backend by @mudler in feat(vibevoice): add new backend #7494
• feat(ui): allow to order search results by @mudler in feat(ui): allow to order search results #7507
• feat(loader): enhance single active backend to support LRU eviction by @mudler in feat(loader): enhance single active backend to support LRU eviction #7535
• feat(stablediffusion-ggml): add lora support by @mudler in feat(stablediffusion-ggml): add lora support #7542
• feat(ui): add mask to install custom backends by @mudler in feat(ui): add mask to install custom backends #7559
• feat(watchdog): add Memory resource reclaimer by @mudler in feat(watchdog): add Memory resource reclaimer #7583
• feat(mlx): add thread-safe LRU prompt cache and min_p/top_k sampling by @blightbow in feat(mlx): add thread-safe LRU prompt cache and min_p/top_k sampling #7556
• feat(whisper): Add prompt to condition transcription output by @richiejp in feat(whisper): Add prompt to condition transcription output #7624

🧠 Models

• feat(stablediffusion): Passthrough more parameters to support z-image and flux2 by @richiejp in feat(stablediffusion): Passthrough more parameters to support z-image and flux2 #7414
• Revert "feat(stablediffusion): Passthrough more parameters to support z-image and flux2" by @mudler in Revert "feat(stablediffusion): Passthrough more parameters to support z-image and flux2" #7417
• feat(stablediffusion): Passthrough more parameters to support z-image and flux2 by @richiejp in feat(stablediffusion): Passthrough more parameters to support z-image and flux2 #7419
• chore(model-gallery): ⬆️ update checksum by @localai-bot in chore(model-gallery): ⬆️ update checksum #7433
• fix(stablediffusion-ggml): Correct Z-Image model name by @richiejp in fix(stablediffusion-ggml): Correct Z-Image model name #7436
• chore(model-gallery): ⬆️ update checksum by @localai-bot in chore(model-gallery): ⬆️ update checksum #7437
• chore(model-gallery): ⬆️ update checksum by @localai-bot in chore(model-gallery): ⬆️ update checksum #7530
• chore(model gallery): 🤖 add 1 new models via gallery agent by @localai-bot in chore(model gallery): 🤖 add 1 new models via gallery agent #7700
• chore(model gallery): 🤖 add 1 new models via gallery agent by @localai-bot in chore(model gallery): 🤖 add 1 new models via gallery agent #7707
• chore(model gallery): 🤖 add 1 new models via gallery agent by @localai-bot in chore(model gallery): 🤖 add 1 new models via gallery agent #7712

📖 Documentation and examples

• chore: Add AGENTS.md by @richiejp in chore: Add AGENTS.md #7688
• chore(deps): Bump llama.cpp to '5b6c9bc0f3c8f55598b9999b65aff7ce4119bc15' and refactor usage of base params by @mudler in chore(deps): Bump llama.cpp to '5b6c9bc0f3c8f55598b9999b65aff7ce4119bc15' and refactor usage of base params #7706

👒 Dependencies

• chore(deps): bump github.com/google/go-containerregistry from 0.19.2 to 0.20.7 by @dependabot[bot] in chore(deps): bump github.com/google/go-containerregistry from 0.19.2 to 0.20.7 #7409
• chore(deps): bump appleboy/ssh-action from 1.2.3 to 1.2.4 by @dependabot[bot] in chore(deps): bump appleboy/ssh-action from 1.2.3 to 1.2.4 #7410
• chore(deps): bump go.opentelemetry.io/otel/sdk/metric from 1.38.0 to 1.39.0 by @dependabot[bot] in chore(deps): bump go.opentelemetry.io/otel/sdk/metric from 1.38.0 to 1.39.0 #7476
• chore(deps): bump actions/stale from 10.1.0 to 10.1.1 by @dependabot[bot] in chore(deps): bump actions/stale from 10.1.0 to 10.1.1 #7473
• chore(deps): bump protobuf from 6.33.1 to 6.33.2 in /backend/python/transformers by @dependabot[bot] in chore(deps): bump protobuf from 6.33.1 to 6.33.2 in /backend/python/transformers #7481
• chore(deps): bump github.com/mudler/cogito from 0.5.1 to 0.6.0 by @dependabot[bot] in chore(deps): bump github.com/mudler/cogito from 0.5.1 to 0.6.0 #7474
• chore(deps): bump go.opentelemetry.io/otel/exporters/prometheus from 0.60.0 to 0.61.0 by @dependabot[bot] in chore(deps): bump go.opentelemetry.io/otel/exporters/prometheus from 0.60.0 to 0.61.0 #7477
• chore(deps): bump github.com/onsi/gomega from 1.38.2 to 1.38.3 by @dependabot[bot] in chore(deps): bump github.com/onsi/gomega from 1.38.2 to 1.38.3 #7475
• chore(deps): bump stable-diffusion.cpp to '8823dc48bcc1598eb9671da7b69e45338d0cc5a5' by @mudler in chore(deps): bump stable-diffusion.cpp to '8823dc48bcc1598eb9671da7b69e45338d0cc5a5' #7524
• chore(makefile): Add buildargs for sd and cuda when building backend by @richiejp in chore(makefile): Add buildargs for sd and cuda when building backend #7525
• chore(deps): bump torch from 2.5.1+cxx11.abi to 2.7.1+cpu in /backend/python/diffusers in the pip group across 1 directory by @dependabot[bot] in chore(deps): bump torch from 2.5.1+cxx11.abi to 2.7.1+cpu in /backend/python/diffusers in the pip group across 1 directory #7549
• chore(deps): bump actions/upload-artifact from 5 to 6 by @dependabot[bot] in chore(deps): bump actions/upload-artifact from 5 to 6 #7585
• chore(deps): bump github.com/onsi/ginkgo/v2 from 2.27.2 to 2.27.3 by @dependabot[bot] in chore(deps): bump github.com/onsi/ginkgo/v2 from 2.27.2 to 2.27.3 #7590
• chore(deps): bump peter-evans/create-pull-request from 7 to 8 by @dependabot[bot] in chore(deps): bump peter-evans/create-pull-request from 7 to 8 #7586
• chore(deps): bump actions/download-artifact from 6 to 7 by @dependabot[bot] in chore(deps): bump actions/download-artifact from 6 to 7 #7587
• chore(deps): bump securego/gosec from 2.22.9 to 2.22.11 by @dependabot[bot] in chore(deps): bump securego/gosec from 2.22.9 to 2.22.11 #7588
• chore(deps): bump github.com/labstack/echo/v4 from 4.13.4 to 4.14.0 by @dependabot[bot] in chore(deps): bump github.com/labstack/echo/v4 from 4.13.4 to 4.14.0 #7589
• chore(deps): bump github.com/jaypipes/ghw from 0.20.0 to 0.21.1 by @dependabot[bot] in chore(deps): bump github.com/jaypipes/ghw from 0.20.0 to 0.21.1 #7591
• chore(deps): bump sentence-transformers from 5.1.0 to 5.2.0 in /backend/python/transformers by @dependabot[bot] in chore(deps): bump sentence-transformers from 5.1.0 to 5.2.0 in /backend/python/transformers #7594
• chore(memory detection): do not use go-sigar as requires CGO on darwin by @mudler in chore(memory detection): do not use go-sigar as requires CGO on darwin #7618
• chore(deps): bump cogito to latest and adapt API changes by @mudler in chore(deps): bump cogito to latest and adapt API changes #7655
• chore(refactor): move logging to common package based on slog by @mudler in chore(refactor): move logging to common package based on slog #7668
• chore(deps): bump xlog to v0.0.3 by @mudler in chore(deps): bump xlog to v0.0.3 #7675
• chore(deps): bump securego/gosec from 2.22.9 to 2.22.11 by @dependabot[bot] in chore(deps): bump securego/gosec from 2.22.9 to 2.22.11 #7690
• chore(deps): bump github.com/mudler/xlog from 0.0.3 to 0.0.4 by @dependabot[bot] in chore(deps): bump github.com/mudler/xlog from 0.0.3 to 0.0.4 #7695
• chore(deps): bump github.com/mudler/cogito from 0.7.1 to 0.7.2 by @dependabot[bot] in chore(deps): bump github.com/mudler/cogito from 0.7.1 to 0.7.2 #7691
• chore(deps): bump github.com/jaypipes/ghw from 0.21.1 to 0.21.2 by @dependabot[bot] in chore(deps): bump github.com/jaypipes/ghw from 0.21.1 to 0.21.2 #7694
• chore(deps): bump github.com/containerd/containerd from 1.7.29 to 1.7.30 by @dependabot[bot] in chore(deps): bump github.com/containerd/containerd from 1.7.29 to 1.7.30 #7692

Other Changes

• chore: ⬆️ Update ggml-org/llama.cpp to eec1e33a9ed71b79422e39cc489719cf4f8e0777 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to eec1e33a9ed71b79422e39cc489719cf4f8e0777 #7363
• Initialize sudo reference before its first actual use by @poretsky in Initialize sudo reference before its first actual use #7367
• chore(deps): update diffusers dependency to use GitHub repo for l4t by @mudler in chore(deps): update diffusers dependency to use GitHub repo for l4t #7369
• chore(l4t/diffusers): bump nvidia l4t index for pytorch 2.9 by @mudler in chore(l4t/diffusers): bump nvidia l4t index for pytorch 2.9 #7379
• Conventional way of adding extra apt repository by @poretsky in Conventional way of adding extra apt repository #7362
• Correct user deletion with all its data by @poretsky in Correct user deletion with all its data #7368
• Clean data directory by @poretsky in Clean data directory #7378
• docs: ⬆️ update docs version mudler/LocalAI by @localai-bot in docs: ⬆️ update docs version mudler/LocalAI #7381
• chore(l4t): Update extra index URL for requirements-l4t.txt by @mudler in chore(l4t): Update extra index URL for requirements-l4t.txt #7383
• chore: Add Python 3.12 support for l4t build profile by @mudler in chore: Add Python 3.12 support for l4t build profile #7384
• chore: ⬆️ Update ggml-org/llama.cpp to 4abef75f2cf2eee75eb5083b30a94cf981587394 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 4abef75f2cf2eee75eb5083b30a94cf981587394 #7382
• chore(diffusers): Add PY_STANDALONE_TAG for l4t Python version by @mudler in chore(diffusers): Add PY_STANDALONE_TAG for l4t Python version #7387
• Revert "chore(l4t): Update extra index URL for requirements-l4t.txt" by @mudler in Revert "chore(l4t): Update extra index URL for requirements-l4t.txt" #7388
• chore: drop pinning of python 3.12 by @mudler in chore: drop pinning of python 3.12 #7389
• chore(deps): bump llama.cpp to 'd82b7a7c1d73c0674698d9601b1bbb0200933f29' by @mudler in chore(deps): bump llama.cpp to 'd82b7a7c1d73c0674698d9601b1bbb0200933f29' #7392
• feat(swagger): update swagger by @localai-bot in feat(swagger): update swagger #7394
• chore: ⬆️ Update ggml-org/llama.cpp to 8c32d9d96d9ae345a0150cae8572859e9aafea0b by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 8c32d9d96d9ae345a0150cae8572859e9aafea0b #7395
• feat(swagger): update swagger by @localai-bot in feat(swagger): update swagger #7400
• chore: ⬆️ Update ggml-org/llama.cpp to 7f8ef50cce40e3e7e4526a3696cb45658190e69a by @mudler in chore: ⬆️ Update ggml-org/llama.cpp to 7f8ef50cce40e3e7e4526a3696cb45658190e69a #7402
• chore: ⬆️ Update ggml-org/llama.cpp to ec18edfcba94dacb166e6523612fc0129cead67a by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to ec18edfcba94dacb166e6523612fc0129cead67a #7406
• chore(deps/stable-diffusion-ggml): update stablediffusion-ggml by @richiejp in chore(deps/stable-diffusion-ggml): update stablediffusion-ggml #7411
• Add Dockerfile for arm64 with nvpl installation by @mudler in Add Dockerfile for arm64 with nvpl installation #7416
• chore: ⬆️ Update ggml-org/llama.cpp to 61bde8e21f4a1f9a98c9205831ca3e55457b4c78 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 61bde8e21f4a1f9a98c9205831ca3e55457b4c78 #7415
• chore: ⬆️ Update leejet/stable-diffusion.cpp to 5865b5e7034801af1a288a9584631730b25272c6 by @localai-bot in chore: ⬆️ Update leejet/stable-diffusion.cpp to 5865b5e7034801af1a288a9584631730b25272c6 #7422
• Messages output fix by @poretsky in Messages output fix #7424
• chore: ⬆️ Update ggml-org/llama.cpp to e9f9483464e6f01d843d7f0293bd9c7bc6b2221c by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to e9f9483464e6f01d843d7f0293bd9c7bc6b2221c #7421
• chore(ui): uniform buttons by @mudler in chore(ui): uniform buttons #7429
• chore(deps): bump llama.cpp to 'bde188d60f58012ada0725c6dd5ba7c69fe4dd87' by @mudler in chore(deps): bump llama.cpp to 'bde188d60f58012ada0725c6dd5ba7c69fe4dd87' #7434
• chore: ⬆️ Update ggml-org/llama.cpp to 8160b38a5fa8a25490ca33ffdd200cda51405688 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 8160b38a5fa8a25490ca33ffdd200cda51405688 #7438
• chore: ⬆️ Update ggml-org/whisper.cpp to a88b93f85f08fc6045e5d8a8c3f94b7be0ac8bce by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to a88b93f85f08fc6045e5d8a8c3f94b7be0ac8bce #7448
• chore: ⬆️ Update ggml-org/llama.cpp to db97837385edfbc772230debbd49e5efae843a71 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to db97837385edfbc772230debbd49e5efae843a71 #7447
• chore(gallery agent): summary now is at root of the git repository by @mudler in chore(gallery agent): summary now is at root of the git repository #7463
• chore(gallery agent): strip thinking tags by @mudler in chore(gallery agent): strip thinking tags #7464
• chore: ⬆️ Update ggml-org/whisper.cpp to a8f45ab11d6731e591ae3d0230be3fec6c2efc91 by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to a8f45ab11d6731e591ae3d0230be3fec6c2efc91 #7483
• chore(deps/llama-cpp): bump to '2fa51c19b028180b35d316e9ed06f5f0f7ada2c1' by @mudler in chore(deps/llama-cpp): bump to '2fa51c19b028180b35d316e9ed06f5f0f7ada2c1' #7484
• chore: ⬆️ Update ggml-org/llama.cpp to 086a63e3a5d2dbbb7183a74db453459e544eb55a by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 086a63e3a5d2dbbb7183a74db453459e544eb55a #7496
• chore: ⬆️ Update ggml-org/whisper.cpp to 9f5ed26e43c680bece09df7bdc8c1b7835f0e537 by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to 9f5ed26e43c680bece09df7bdc8c1b7835f0e537 #7509
• chore: ⬆️ Update ggml-org/llama.cpp to 4dff236a522bd0ed949331d6cb1ee2a1b3615c35 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 4dff236a522bd0ed949331d6cb1ee2a1b3615c35 #7508
• chore: ⬆️ Update ggml-org/llama.cpp to a81a569577cc38b32558958b048228150be63eae by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to a81a569577cc38b32558958b048228150be63eae #7529
• chore: ⬆️ Update leejet/stable-diffusion.cpp to 11ab095230b2b67210f5da4d901588d56c71fe3a by @localai-bot in chore: ⬆️ Update leejet/stable-diffusion.cpp to 11ab095230b2b67210f5da4d901588d56c71fe3a #7539
• chore(l4t13): use pytorch index by @mudler in chore(l4t13): use pytorch index #7546
• Revert "chore(deps): bump torch from 2.5.1+cxx11.abi to 2.7.1+cpu in /backend/python/diffusers in the pip group across 1 directory" by @mudler in Revert "chore(deps): bump torch from 2.5.1+cxx11.abi to 2.7.1+cpu in /backend/python/diffusers in the pip group across 1 directory" #7558
• chore: ⬆️ Update ggml-org/whisper.cpp to 2551e4ce98db69027d08bd99bcc3f1a4e2ad2cef by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to 2551e4ce98db69027d08bd99bcc3f1a4e2ad2cef #7561
• chore: ⬆️ Update leejet/stable-diffusion.cpp to 43a70e819b9254dee0d017305d6992f6bb27f850 by @localai-bot in chore: ⬆️ Update leejet/stable-diffusion.cpp to 43a70e819b9254dee0d017305d6992f6bb27f850 #7562
• chore: ⬆️ Update ggml-org/llama.cpp to 5266379bcae74214af397f36aa81b2a08b15d545 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 5266379bcae74214af397f36aa81b2a08b15d545 #7563
• chore: ⬆️ Update ggml-org/llama.cpp to 5c8a717128cc98aa9e5b1c44652f5cf458fd426e by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 5c8a717128cc98aa9e5b1c44652f5cf458fd426e #7573
• chore(llama.cpp): Add Missing llama.cpp Options to gRPC Server by @mudler in chore(llama.cpp): Add Missing llama.cpp Options to gRPC Server #7584
• chore: ⬆️ Update leejet/stable-diffusion.cpp to 200cb6f2ca07e40fa83b610a4e595f4da06ec709 by @localai-bot in chore: ⬆️ Update leejet/stable-diffusion.cpp to 200cb6f2ca07e40fa83b610a4e595f4da06ec709 #7597
• Revert "chore(deps): bump securego/gosec from 2.22.9 to 2.22.11" by @mudler in Revert "chore(deps): bump securego/gosec from 2.22.9 to 2.22.11" #7602
• chore: ⬆️ Update ggml-org/llama.cpp to ef83fb8601229ff650d952985be47e82d644bfaa by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to ef83fb8601229ff650d952985be47e82d644bfaa #7611
• chore: ⬆️ Update ggml-org/whisper.cpp to 3e79e73eee32e924fbd34587f2f2ac5a45a26b61 by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to 3e79e73eee32e924fbd34587f2f2ac5a45a26b61 #7630
• chore: ⬆️ Update ggml-org/llama.cpp to d37fc935059211454e9ad2e2a44e8ed78fd6d1ce by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to d37fc935059211454e9ad2e2a44e8ed78fd6d1ce #7629
• chore: ⬆️ Update leejet/stable-diffusion.cpp to bda7fab9f208dff4b67179a68f694b6ddec13326 by @richiejp in chore: ⬆️ Update leejet/stable-diffusion.cpp to bda7fab9f208dff4b67179a68f694b6ddec13326 #7639
• chore: ⬆️ Update ggml-org/llama.cpp to f9ec8858edea4a0ecfea149d6815ebfb5ecc3bcd by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to f9ec8858edea4a0ecfea149d6815ebfb5ecc3bcd #7642
• chore: ⬆️ Update ggml-org/whisper.cpp to 6c22e792cb0ee155b6587ce71a8410c3aeb06949 by @localai-bot in chore: ⬆️ Update ggml-org/whisper.cpp to 6c22e792cb0ee155b6587ce71a8410c3aeb06949 #7644
• chore: ⬆️ Update ggml-org/llama.cpp to ce734a8a2f9fb6eb4f0383ab1370a1b0014ab787 by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to ce734a8a2f9fb6eb4f0383ab1370a1b0014ab787 #7654
• chore(cogito): respect application-level logging and propagate by @mudler in chore(cogito): respect application-level logging and propagate #7656
• chore: ⬆️ Update ggml-org/llama.cpp to 52ab19df633f3de5d4db171a16f2d9edd2342fec by @localai-bot in chore: ⬆️ Update ggml-org/llama.cpp to 52ab19df633f3de5d4db171a16f2d9edd2342fec #7665
• docs: Add langchain-localai integration package to documentation by @mkhludnev in docs: Add langchain-localai integration package to documentation #7677
• chore: allow to set local-ai log format, default to custom one by @mudler in chore: allow to set local-ai log format, default to custom one #7679
• chore(deps): bump llama.cpp to '0e1ccf15c7b6d05c720551b537857ecf6194d420' by @mudler in chore(deps): bump llama.cpp to '0e1ccf15c7b6d05c720551b537857ecf6194d420' #7684
• chore(gallery agent): various fixups by @mudler in chore(gallery agent): various fixups #7697
• Revert "chore(deps): bump securego/gosec from 2.22.9 to 2.22.11" by @mudler in Revert "chore(deps): bump securego/gosec from 2.22.9 to 2.22.11" #7698
• chore(logging): be consistent and do not emit logs from echo by @mudler in chore(logging): be consistent and do not emit logs from echo #7710

New Contributors

• @rampa3 made their first contribution in fix(ui): Update few links in web UI from 'browse' to '/browse/' #7445
• @blightbow made their first contribution in feat(mlx): add thread-safe LRU prompt cache and min_p/top_k sampling #7556

Full Changelog: v3.8.0...v3.9.0

---

This discussion was created from the release v3.9.0.

[Clawue884]

Hi LocalAI team 👋
Great work on LocalAI 3.9.0 — Agent Jobs Panel and the new Smart Memory Reclaimer are really impressive.
Quick question regarding Agent Jobs + LRU model eviction under concurrent workloads:
If multiple Agent Jobs run at the same time and memory pressure triggers LRU eviction:
Are models currently used by active jobs protected from eviction?
Or can a model be unloaded while a job is still running?
Understanding this would help when designing scheduled or long-running agent workflows on limited VRAM systems.
Thanks again for the great release and focus on stability 🙏