#Open-WebUI
A six-year-old NVIDIA RTX 3090, combined with Linux Mint, Ollama and Open WebUI, proved capable of running GPT-OSS 20B entirely locally at an impressive 144 tokens per second. Offering greater privacy, control, customization & freedom from cloud inference fees... p4sc4l.substack.com/p/a-six-year...
A six-year-old NVIDIA RTX 3090, combined with Linux Mint, Ollama and Open WebUI, proved capable of running GPT-OSS 20B entirely locally at an impressive 144 tokens per second.
Local AI offers greater privacy, control, customization & freedom from cloud inference fees, although hardware requirements, maintenance, security & model reliability remain important considerations.
p4sc4l.substack.com
November 15, 2026 at 10:26 PM
Lo de Open WebUI no tiene ningún sentido, cómo pueden tomar siempre LA PEOR decisión de diseño en cada feature? Es que wtf, parece que lo hagan a propósito.
November 13, 2026 at 11:24 AM
Adding EastX to Open WebUI? Configure the OpenAI-compatible connection as an admin and check server egress, exact model IDs and duplicate names.

Guide: https://eastx.ai/zh/community/open-webui-eastx-setup
November 15, 2026 at 5:56 AM
open-webui-tether-bin
Desktop companion for Open WebUI
aur.archlinux.org
November 11, 2026 at 10:08 PM
Open WebUI with SearXNG web search and local Ollama models #KadetHub #Kapitan #OpenWebUI #SearXNG #Ollama
Open WebUI with SearXNG web search and local Ollama models
Open WebUI with SearXNG as its web search and Ollama on Kubernetes, from four Kapitan recipes, deployed on kind and tested through Open WebUI's API.
kadet-hub.org
November 10, 2026 at 10:30 PM
One Second: The Open WebUI Windows Master Key https://packetstorm.news/files/241258 #paper
November 14, 2026 at 12:27 AM
Museに一度弾かれると画像の調整が一切合切出来なくなるから、そうなったらOpen WebUIからGPT APIを叩いてGPT image使って出力させられる!
もちろん、エッチなのはダメ!死刑!!
November 13, 2026 at 9:33 AM
Does setting function calling to LEGACY still let tool use work in Open WebUI with those ollama models or does it just fall back to prompt based calling
November 10, 2026 at 9:21 PM
📰 Open WebUI v0.13.1: Coding Agents in the Model Picker

Inside:
• Tether Connects Your Computer to Open WebUI Without Open Ports

Full breakdown, every claim linked to its source 👇
Open WebUI v0.13.1: Coding Agents in the Model Picker
Open WebUI v0.13.1 adds coding agents to the model picker, the Tether desktop app and an Auto tool permission mode, and changes RAG defaults.
thestacksentinel.com
November 8, 2026 at 10:30 PM
OPEN WEBUI V0.13.3: CODING AGENTS IN AUTOMATIONS, TERMINAL AND FOLDER SWITCHING DURING CHAT
November 9, 2026 at 12:30 AM
Si te cansa Open WebUI, prueba LibreChat: más sobrio, multiproveedor y la configuración vive en un yaml, así no dependes de cada cambio de la interfaz. Para chatear con documentos, AnythingLLM va bien.
November 14, 2026 at 8:11 AM
というわけで、火を入れてから2~3時間で、Open WebUI+ローカルLLMでチャットができるようになりました。もちろん日本語で😀
macOS(というよりUNIX)なら、すぐにローカルAIをはじめられます🍎
November 16, 2026 at 8:57 AM
A six-year-old NVIDIA RTX 3090, combined with Linux Mint, Ollama and Open WebUI, proved capable of running GPT-OSS 20B entirely locally at an impressive 144 tokens per second. Offering greater privacy, control, customization & freedom from cloud inference fees... p4sc4l.substack.com/p/a-six-year...
A six-year-old NVIDIA RTX 3090, combined with Linux Mint, Ollama and Open WebUI, proved capable of running GPT-OSS 20B entirely locally at an impressive 144 tokens per second.
Local AI offers greater privacy, control, customization & freedom from cloud inference fees, although hardware requirements, maintenance, security & model reliability remain important considerations.
p4sc4l.substack.com
November 15, 2026 at 10:25 PM
As far as I know Legacy keeps tools working, just through prompting: Open WebUI puts the tool list in the prompt and parses the reply, so the model doesn't need native tool-call support. Fine for models without it, a bit less reliable with tiny ones.
November 11, 2026 at 8:01 AM
Open WebUI v0.13.1 is out for the open-source AI workspace ✨ It adds coding agents (Claude Code, Codex) to the model picker and the Tether desktop app. Soon on PikaPods.

go.pf7.net/WUwQu
go.pf7.net/srmBf #opensource #ownyourdata #datasovereignty #techfreedom
November 8, 2026 at 6:53 PM
Open WebUI: the Windows installer mints the master key from the install second — CVE-2025-15603 was rejected; here is the measurement

Posted by theknell--- via Fulldisclosure on Nov 12Open WebUI's Windows installer (start_windows.bat) generates
WEBUI_SECRET_KEY — the master secr…
#hackernews #news
Open WebUI: the Windows installer mints the master key from the install second — CVE-2025-15603 was rejected; here is the measurement
Posted by theknell--- via Fulldisclosure on Nov 12Open WebUI's Windows installer (start_windows.bat) generates WEBUI_SECRET_KEY — the master secret that signs every session token — from the C library's rand(), a linear congruential generator seeded once with the current time in seconds. The key's entire entropy is the second in which the software was installed. From any ordinary user's own token, the master key is recoverable blind in ~2.1 seconds across a ±1-day...
seclists.org
November 14, 2026 at 4:01 AM
SpecStream now answers SearXNG's JSON at one address, on the most accurate web search API.

Paste https://api.specstream.ai/searxng/YOUR_KEY/search into Open WebUI's Searxng Query URL and web search works with no SearXNG server to keep running.

$5 per 1,000 searches.
Open WebUI web search setup: one URL and one key
Set up web search in Open WebUI with one address. SpecStream answers SearXNG's JSON and Open WebUI's external search at $5 per 1,000 searches.
specstream.ai
November 10, 2026 at 7:24 AM
OPEN WEBUI V0.13.1: CODING AGENTS AS MODELS, TETHER DESKTOP APP
November 8, 2026 at 11:30 AM