39 Pull requests merged by 3 people

• chore: ⬆️ Update ggml-org/llama.cpp to 814f795e063c257f33b921eab4073484238a151a

  #5331 merged May 8, 2025

• chore(model gallery): add qwen3-14b-griffon-i1

  #5330 merged May 7, 2025

• chore: ⬆️ Update ggml-org/llama.cpp to 91a86a6f354aa73a7aab7bc3d283be410fdc93a5

  #5329 merged May 6, 2025

• chore(deps): bump llama.cpp to b34c859146630dff136943abc9852ca173a7c9d6

  #5323 merged May 6, 2025

• chore(model gallery): add claria-14b

  #5326 merged May 6, 2025

• chore(model gallery): add goekdeniz-guelmez_josiefied-qwen3-8b-abliterated-v1

  #5325 merged May 6, 2025

• chore(model gallery): add huihui-ai_qwen3-14b-abliterated

  #5324 merged May 6, 2025

• chore(model-gallery): ⬆️ update checksum

  #5321 merged May 6, 2025

• fix(hipblas): do not build all cpu-specific flags

  #5322 merged May 6, 2025

• chore(deps): bump mxschmitt/action-tmate from 3.21 to 3.22

  #5319 merged May 5, 2025

• fix(arm64): do not build instructions which are not available

  #5318 merged May 5, 2025

• chore(model gallery): add allura-org_remnant-qwen3-8b

  #5317 merged May 5, 2025

• chore: ⬆️ Update ggml-org/llama.cpp to 9fdfcdaeddd1ef57c6d041b89cd8fb7048a0f028

  #5316 merged May 4, 2025

• fix: use rice when embedding large binaries

  #5309 merged May 4, 2025

• chore(model gallery): add rei-v3-kto-12b

  #5313 merged May 4, 2025

• chore(model gallery): add kalomaze_qwen3-16b-a3b

  #5312 merged May 4, 2025

• chore(model gallery): add qwen3-30b-a1.5b-high-speed

  #5311 merged May 4, 2025

• chore: ⬆️ Update ggml-org/llama.cpp to 36667c8edcded08063ed51c7d57e9e086bbfc903

  #5300 merged May 4, 2025

• fix(gpu): do not assume gpu being returned has node and mem

  #5310 merged May 3, 2025

• chore(defaults): enlarge defaults, drop gpu layers which is infered

  #5308 merged May 3, 2025

• chore(deps): bump llama.cpp to '1d36b3670b285e69e58b9d687c770a2a0a192194

  #5307 merged May 3, 2025

• chore(model gallery): add smoothie-qwen3-8b

  #5306 merged May 3, 2025

• chore(model gallery): add qwen-3-32b-medical-reasoning-i1

  #5305 merged May 3, 2025

• chore(model gallery): add amoral-qwen3-14b

  #5304 merged May 3, 2025

• chore(model gallery): add comet_12b_v.5-i1

  #5303 merged May 3, 2025

• chore(model gallery): add genericrpv3-4b

  #5302 merged May 3, 2025

• chore(model gallery): add planetoid_27b_v.2

  #5301 merged May 3, 2025

• feat(llama.cpp): estimate vram usage

  #5299 merged May 2, 2025

• chore(model gallery): add webthinker-qwq-32b-i1

  #5298 merged May 2, 2025

• chore(model gallery): add shuttleai_shuttle-3.5

  #5297 merged May 2, 2025

• chore(model gallery): add microsoft_phi-4-reasoning

  #5296 merged May 2, 2025

• chore(model gallery): add microsoft_phi-4-reasoning-plus

  #5295 merged May 2, 2025

• chore(model gallery): add furina-8b

  #5294 merged May 2, 2025

• chore(model gallery): add josiefied-qwen3-8b-abliterated-v1

  #5293 merged May 2, 2025

• chore: ⬆️ Update ggml-org/llama.cpp to d7a14c42a1883a34a6553cbfe30da1e1b84dfd6a

  #5292 merged May 2, 2025

• chore(model gallery): add microsoft_phi-4-mini-reasoning

  #5288 merged May 1, 2025

• chore(model gallery): add fast-math-qwen3-14b

  #5287 merged May 1, 2025

• chore(model gallery): add qwen3-8b-jailbroken

  #5286 merged May 1, 2025

• chore(model gallery): add qwen3-30b-a3b-abliterated

  #5285 merged May 1, 2025

3 Issues closed by 2 people

• feat: automatically adjust default gpu_layers by available GPU memory

  #3541 closed May 2, 2025

• feat: Improve automatic VRAM allocation from llama-cpp

  #5257 closed May 2, 2025

• Issues running a few Models.

  #4231 closed May 1, 2025

7 Issues opened by 6 people

• Possible Deadlock on loading model with LOCALAI_SINGLE_ACTIVE_BACKEND=true

  #5328 opened May 6, 2025

• Whisper API not compatible with OpenAI

  #5327 opened May 6, 2025

• Backend GGML outdated builds causing it to fail on newer nvidia GPUs

  #5315 opened May 4, 2025

• Model fails to load with RCP error after clean install

  #5314 opened May 4, 2025

• error with txt2vid example in docs

  #5291 opened May 1, 2025

• model used in img2vid example in docs is no longer available/valid

  #5290 opened May 1, 2025

• [Bug] Multilingual-e5-base Embeddings Issue with llama-embeddings Backend on CUDA 12 Docker (Windows 11)

  #5289 opened May 1, 2025

8 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Completion endpoint does not count tokens when using vLLM backend

  #3436 commented on May 1, 2025 • 0 new comments

• Watchdog Time Out detection error.

  #5221 commented on May 2, 2025 • 0 new comments

• LocalAI v2.28 does not use the ENV file in docker.

  #5280 commented on May 4, 2025 • 0 new comments

• Clean install fails to run any model

  #5225 commented on May 5, 2025 • 0 new comments

• Installation problems on MacOS

  #4300 commented on May 5, 2025 • 0 new comments

• Can't build LocalAI with llama.cpp with CUDA

  #3418 commented on May 6, 2025 • 0 new comments

• chore: :arrow_up: Update PABannier/bark.cpp to `5d5be84f089ab9ea53b7a793f088d3fbf7247495`

  #4786 commented on May 7, 2025 • 0 new comments

• chore: :arrow_up: Update leejet/stable-diffusion.cpp to `10c6501bd05a697e014f1bee3a84e5664290c489`

  #4925 commented on May 7, 2025 • 0 new comments