I’ll answer it since he won’t lol
I have a Xeon whatever with integrated graphics, no real gpu. I think at least 64gb ram too. Running ollama qwen3:4b with open web ui and it seems connected and working through Umbrel is 1.7.4 configurations. It takes 1-5 minutes per response. The responses are blank. It’s logging token use and chat history and even auto summarizing chat titles but I have yet to get any response from it besides something like {parameter}{date}{time} some jargon crap like that. So far I am wholly unimpressed with “local, home, self-hosted AI”. Completely useless.
npub1ng…gckc7
2026-09-12 10:43:25 UTC
in reply to nevent1q…pfde
Author Public Key
npub1ngzq23mhwam8xqkntj5zv7686fj9vnknkh85l5c59wrs2j8jj8uq9gckc7Published at
2026-09-12 10:43:25 UTCEvent JSON
{
"id": "00000081f2981af87c98bd41b6e270926c3876fed1996791747463d75ff9aa77",
"pubkey": "9a0405477777767302d35ca8267b47d264564ed3b5cf4fd3142b870548f291f8",
"created_at": 1789209805,
"kind": 1,
"tags": [
[
"e",
"00002746fe74efd69e34ad18feceec6160365f7c8762c9a9ac6f51dbfac401fb",
"",
"root"
],
[
"p",
"e2ccf7cf20403f3f2a4a55b328f0de3be38558a7d5f33632fdaaefc726c1c8eb"
],
[
"p",
"9a0405477777767302d35ca8267b47d264564ed3b5cf4fd3142b870548f291f8"
],
[
"client",
"Wisp iOS"
],
[
"nonce",
"54182",
"16"
]
],
"content": "I’ll answer it since he won’t lol \n\nI have a Xeon whatever with integrated graphics, no real gpu. I think at least 64gb ram too. Running ollama qwen3:4b with open web ui and it seems connected and working through Umbrel is 1.7.4 configurations. It takes 1-5 minutes per response. The responses are blank. It’s logging token use and chat history and even auto summarizing chat titles but I have yet to get any response from it besides something like {parameter}{date}{time} some jargon crap like that. So far I am wholly unimpressed with “local, home, self-hosted AI”. Completely useless.",
"sig": "7053034946df9f1b7e8ee7a811730693003d4e6c63dfe14eb201a874ba3f45b44368a1fc625f27284133f83a4db22d3a34e22634b15f69e4393521c40da6a170"
}