v0.20.4 - Update Ollama to v0.20.4 - Update Open WebUI to v0.8.12 - 2026-04-08

v0.20.3 - Update Ollama to v0.20.3 - Update Open WebUI to v0.8.12 - 2026-04-07

v0.17.3 - Update Ollama to v0.17.3 - Update Open WebUI to v0.8.5 - 2026-02-23

v0.9.11 - Upgrade Ollama to the latest 2.3.0-SNAPSHOT (0.9.3) version with Intel support - Upgrade Open-Webui to the latest main branch version

0. The problem of slow loading of the model for the first time is still under optimization... 1. Open-Webui uses the built-in qwen 2.5:0.5b as a TaskModel 2. Open-Webui disables the OPENAI API by default to avoid requesting every query 3. Unzip the Ipex oneAPI driver under the persistence directory 4. Open-WeBui includes one sentence-transformers/all-MiniLM-L6-v2 to avoid starting and downloading for too long 5. Open-WeBui disables automatic updates of RAG-related models by default to avoid requesting stuck every time it is launched