Issues · mudler/LocalAI · GitHub

Labels Milestones New issue

Traces are lost when restarting the service

#9308

· pmarini-nc opened

on Apr 10, 2026

Memory RAG fails to upload PDF – "model granite-embedding-107m-multilingual does not exist" when using custom embedding model

#9307

· zsj1029 opened

on Apr 10, 2026

[whisper] word timings in verbose_json

#9306

· eglia opened

on Apr 10, 2026

Exclude models from idle unloading

#9305

· loryanstrant opened

on Apr 10, 2026

Add toggle to enable/disable models from being able to load.

#9303

· neurocis opened

on Apr 9, 2026

If a model supports multiple backends from the model gallery like qwen-asr, ask us which backends we want to install

#9295

· JohnGalt1717 opened

on Apr 9, 2026

Renaming a model causes duplicates in the UI

#9294

· JohnGalt1717 opened

on Apr 9, 2026

qwen3-tts-1.7b-custom-voice vllm-omni backend map fails. Qwen backend map has errors

#9293

· JohnGalt1717 opened

on Apr 9, 2026

Intel Arc Pro B70 VRAM not parsed

#9289

· eglia opened

on Apr 9, 2026

If you add opus as a backend via api, then add a model via the api that should use llama.cpp or vllm, it then tries to use opus

#9287

· JohnGalt1717 opened

on Apr 9, 2026

llama-cpp-avx2 gRPC backend segfaults loading Qwen3-VL GGUF models

#9279

· rmdevpro opened

on Apr 8, 2026

Vulnerability in LocalAI project

#9277

· ankitdn opened

on Apr 8, 2026