#open-webui — Public Fediverse posts

Nice, just activated #Mistral OCR in my local #OpenWebUI

https://www.heise.de/news/Dokumente-verstehen-statt-nur-lesen-Mistral-OCR-4-ist-da-11343027.html

#MistralOCR

Nice, just activated #Mistral OCR in my local #OpenWebUI

https://www.heise.de/news/Dokumente-verstehen-statt-nur-lesen-Mistral-OCR-4-ist-da-11343027.html

#MistralOCR

CVE-2026-54008 - SSRF in Open WebUI. OAuth picture URL redirect validation bypass. CVSS 8.5. No patch available. Disable OAuth until fixed. #CVE #OpenWebUI #infosec

https://www.valtersit.com/cve/CVE-2026-54008/

It's really "wow" (positively and frightening), how easy voices can be cloned nowadays ... just tried a voice clone for the voice output for my #OpenWebUI and #mistral

It's really "wow" (positively and frightening), how easy voices can be cloned nowadays ... just tried a voice clone for the voice output for my #OpenWebUI and #mistral

📱 cogwheel0/conduit

Mobile (iOS/Android) client for OpenWebUI. Chat with your self‑hosted AI.

A native iOS and Android app enabling chat with self-hosted Open WebUI instances, supporting authentication, streaming, and mobile-optimized features

⭐ Stars: 1693
📅 Last Update: Jun 19, 2026

https://github.com/cogwheel0/conduit

#selfhosted #homelab #selfhost #selfhosting #opensource #mobile #openwebui

Got a local #LLM running, #Gemma4 latest turned out to be the best, #openwebui plus #Ollama.

Got it generating novel-style roleplay responses within a couple of seconds, negligible system resource impact aside from 9GB of VRAM plus about 100% of my GPU whenever it's "thinking".

Fast, local, and inherently private, not too big.

Tried a couple of smaller models; unimpressive. Larger models were too slow.

I think this will work for what I want, though, fleshing out characters for a story.

Got a local #LLM running, #Gemma4 latest turned out to be the best, #openwebui plus #Ollama.

Got it generating novel-style roleplay responses within a couple of seconds, negligible system resource impact aside from 9GB of VRAM plus about 100% of my GPU whenever it's "thinking".

Fast, local, and inherently private, not too big.

Tried a couple of smaller models; unimpressive. Larger models were too slow.

I think this will work for what I want, though, fleshing out characters for a story.

OpenClaw im Team-Chat: Was das Open WebUI Channels Plugin wirklich bringt

Ein Community-Plugin bringt OpenClaw in Open WebUI Channels. Praktisch ist das vor allem dort, wo Teams Agenten direkt im gemeinsamen Chat ansprechen wollen.

https://agentenlog.de/posts/2026-06-09-openclaw-open-webui-channels-team-chat

#openclaw #openwebui

Passend zum #DigitalIndependenceDay: zur #lernOS Convention haben wir den #Chatbot Suri gebaut, der auf der #loscon26 Website unten rechts Auskunft zu Format und Programm geben kann.

Suri ist nicht etwa ein OpenAI CustomGPT, sondern ist mit Low-Code/No-Code komplett selbst gebaut. Die Zutaten sind #OpenWebUI und #n8n.

Ausserdem verwenden wir kein amerikanisches oder chinesisches LLM, sondern mit #Mistral 4 ein europäisches.

Wie das geht zeige ich am 23./24. Juni: https://loscon.lernos.org

Passend zum #DigitalIndependenceDay: zur #lernOS Convention haben wir den #Chatbot Suri gebaut, der auf der #loscon26 Website unten rechts Auskunft zu Format und Programm geben kann.

Suri ist nicht etwa ein OpenAI CustomGPT, sondern ist mit Low-Code/No-Code komplett selbst gebaut. Die Zutaten sind #OpenWebUI und #n8n.

Ausserdem verwenden wir kein amerikanisches oder chinesisches LLM, sondern mit #Mistral 4 ein europäisches.

Wie das geht zeige ich am 23./24. Juni: https://loscon.lernos.org

Как я сделал локальный RAG-сервис для SRE: ищем по документации, ранбукам и коду через Ollama

Недавно делал учебный проект про автоматизацию документирования инцидентов. Поначалу планы были грандиозными: инциденты, таймлайны, интеграции с мониторингами, чатами, постмортемы, подсказки дежурным инженерам. Но довольно быстро стало понятно, что с временными и ресурсными ограничениями лучше не пытаться написать маленький PagerDuty. Поэтому я сузил задачу до более реалистичного ядра: локального RAG-сервиса, который ищет по документации, ранбукам и коду, а затем передаёт найденный контекст в LLM. Так появился llmortem — FastAPI-сервис, который можно подключить к OpenWebUI как OpenAI-compatible backend. В статье расскажу, как устроена архитектура, почему я начал с BM25, зачем индексировать docstring’и и какие ограничения у такого подхода.

https://habr.com/ru/articles/1043928/

#RAG #LLM #FastAPI #Python #Ollama #OpenWebUI #BM25 #SRE #ранбуки #OpenAIcompatible_API

nächster schritt... mein setup verstehen :D ich hab lm studio auf meinem macstudio laufen -> auf meinem #vserver #openwebui - verbunden sind die über #tailscale

da hab ich damals (vor 2 monaten) was eingerichtet was so gut funktioniert hat das ich es gleich wieder vergessen habe

-> nun geh ich von der vm direkt auf beim lm studio ohne den umweg über das internet =)

installation von hermes läuft - ist ne andere hausnimmer wie #opencode

python3.11 und node wird installiert

installtion läuft nun seit über 5 minuten :D

@Emma_Wilson It was not that difficult, the export of #Perplexity is indeed quite good, if not that we loose some important information, like the pair collection UUID-name, the reference links, and I think the model used too... at the time of the last export I only ever used Research (was unlimited at the time) and the normal search... But I need to update my GDPR export now that half a year has passed, and I started using the individual models a lot, so I will find it out relatively soon (1/2 months for the delivery) ^^ At the moment I only have the "mode" which is either "COPILOT" or "CONCISE", not sure if it is the Research vs. Pro Search or something else ^^"

For the collections, we can post-process them, as the UUID is exported, but there's no mapping to the name used on the webapp, so there's some manual digging to do, and I could not get to import the collections to #OpenWebUI without having to use the API, loosing the timings. AI suggested me to go edit the database to achieve it, and I'm sure that we can script a way to move the conversation into the relative folder, for this reason have added an option to add the collction UUID as part of the title, which then could be stripped so it is no longer part of the title, but haven't dug deeper :)

The threads are retained, so I have my query, the answer, and the next query I made, and so on.. also, the creation and update time are provided, so they are timed correctly on the chats list on OWUI ^^

Overall, I am sufficiently happy with the export they've provided me, I only wished I didn't need to appeal to the #GDPR to have it... and having to do it again, and wait months to have the exports 😅 Also, this cuts out anyone else not being an European Citizen, and it's sad..

#Europe

@Emma_Wilson It was not that difficult, the export of #Perplexity is indeed quite good, if not that we loose some important information, like the pair collection UUID-name, the reference links, and I think the model used too... at the time of the last export I only ever used Research (was unlimited at the time) and the normal search... But I need to update my GDPR export now that half a year has passed, and I started using the individual models a lot, so I will find it out relatively soon (1/2 months for the delivery) ^^ At the moment I only have the "mode" which is either "COPILOT" or "CONCISE", not sure if it is the Research vs. Pro Search or something else ^^"

For the collections, we can post-process them, as the UUID is exported, but there's no mapping to the name used on the webapp, so there's some manual digging to do, and I could not get to import the collections to #OpenWebUI without having to use the API, loosing the timings. AI suggested me to go edit the database to achieve it, and I'm sure that we can script a way to move the conversation into the relative folder, for this reason have added an option to add the collction UUID as part of the title, which then could be stripped so it is no longer part of the title, but haven't dug deeper :)

The threads are retained, so I have my query, the answer, and the next query I made, and so on.. also, the creation and update time are provided, so they are timed correctly on the chats list on OWUI ^^

Overall, I am sufficiently happy with the export they've provided me, I only wished I didn't need to appeal to the #GDPR to have it... and having to do it again, and wait months to have the exports 😅 Also, this cuts out anyone else not being an European Citizen, and it's sad..

#Europe

@robux4 +1 👍️

I actually like the fact that my local LLM setups (#Ollama, #openWebUI) on a 12GB GPU are very useful for many tasks already - yet *take some time* to process.

And I can *hear* (the fan noise) if it has to "think" 😉.

The reduced response-time causes one to use it "only when needed or makes sense".

Like a natural self-regulation? 🌻

(yeah, and no hidden costs)

@robux4 +1 👍️

I actually like the fact that my local LLM setups (#Ollama, #openWebUI) on a 12GB GPU are very useful for many tasks already - yet *take some time* to process.

And I can *hear* (the fan noise) if it has to "think" 😉.

The reduced response-time causes one to use it "only when needed or makes sense".

Like a natural self-regulation? 🌻

(yeah, and no hidden costs)

Vibecoded a variant of the previous perplexity-export-convert that converted into markdowns, this one exports #Perplexity chats into a json compatible file that can be imported into #OpenWebUI, so you can go full self-hosting without loosing access to your previous queries ^^

https://github.com/FraYoshi/perplexity-to-openwebui

Hope it will help some of you! ^^
P.s. It's necessary to issue a #GDPR export request first, so I'm afraid it's an #Europe only option..

#SelfHosting #vibecoding #AI #LLM

Vibecoded a variant of the previous perplexity-export-convert that converted into markdowns, this one exports #Perplexity chats into a json compatible file that can be imported into #OpenWebUI, so you can go full self-hosting without loosing access to your previous queries ^^

https://github.com/FraYoshi/perplexity-to-openwebui

Hope it will help some of you! ^^
P.s. It's necessary to issue a #GDPR export request first, so I'm afraid it's an #Europe only option..

#SelfHosting #vibecoding #AI #LLM

Как мы построили корпоративную LLM-платформу: архитектура, грабли и выводы

Обычно внедрение AI в компаниях происходит по такому сценарию: собрали одного ассистента, показали руководству, получили аплодисменты. Потом второго, третьего — и через полгода получается зоопарк с разными ключами, моделями и неконтролируемым бюджетом. Вместо набора разрозненных ассистентов мы сразу пошли в платформу. В статье рассказываем, из чего она состоит, как эволюционировало наше железо, зачем понадобилось два слоя наблюдаемости и почему маркетплейсный RAG ломается на PDF-файлах. С графиками, схемой архитектуры и выводами, которые сами хотели бы прочитать год назад.

https://habr.com/ru/companies/sminex_developer/articles/1037438/

#ai #llm #openwebui #langflow #langfuse #litellm #vllm #openai

Как мы в отделе документации создали LLM агента для автоматизированного перевода с английского на другие языки

Разбираем, как в отделе документации построили LLM-агента для автоматизированного перевода Markdown-документации. Архитектура, пайплайн, валидация, работа с Ollama, OpenWebUI и Qwen, плюсы и ограничения подхода.

https://habr.com/ru/companies/hostkey/articles/1037646/

#LLM #автоматизация_перевода #техническая_документация #Python #валидация #Markdown #OpenWebUI #Qwen #оркестрация #hostkey

Poważna luka w Open WebUI (0.7.2) prowadzi do 1-click RCE. PoC udostępniony przez badacza po zignorowaniu jego zgłoszenia

Czy jedno kliknięcie wystarczy, aby całe środowisko AI wpadło w ręce cyberprzestępców? W przypadku Open WebUI odpowiedź brzmi tak. Jak pokazał badacz bezpieczeństwa Metin Yunus Kandemir, wykryta podatność (Stored XSS) pozwala na przejęcie pełnej kontroli nad platformą przy minimalnej interakcji ze strony użytkownika. TLDR: Błąd występuje w implementacji funkcji przetwarzającej...

#WBiegu #Ai #Exploit #OpenWebui #Poc #Rce #XSS

https://sekurak.pl/powazna-luka-w-open-webui-0-7-2-prowadzi-do-1-click-rce-poc-udostepniony-przez-badacza-po-zignorowaniu-jego-zgloszenia/

Poważna luka w Open WebUI (0.7.2) prowadzi do 1-click RCE. PoC udostępniony przez badacza po zignorowaniu jego zgłoszenia

Czy jedno kliknięcie wystarczy, aby całe środowisko AI wpadło w ręce cyberprzestępców? W przypadku Open WebUI odpowiedź brzmi tak. Jak pokazał badacz bezpieczeństwa Metin Yunus Kandemir, wykryta podatność (Stored XSS) pozwala na przejęcie pełnej kontroli nad platformą przy minimalnej interakcji ze strony użytkownika. TLDR: Błąd występuje w implementacji funkcji przetwarzającej...

#WBiegu #Ai #Exploit #OpenWebui #Poc #Rce #XSS

https://sekurak.pl/powazna-luka-w-open-webui-0-7-2-prowadzi-do-1-click-rce-poc-udostepniony-przez-badacza-po-zignorowaniu-jego-zgloszenia/

RT @Moehrenmann: Und passend dazu hat sich noch dieses niedliche Projekt gefunden, das „wir“ gerade zum Laufen gebracht haben und das Agenten wie Hermes oder Openclaw explizit unterstützt. Minimalistisch und dennoch funktional. github.com/garrytan/gbrain Klar Moehre (@Moehrenmann) Ich spiele jetzt seit einer Woche mit diesem Konzept in @OpenWebUI und es funktioniert erstaunlich gut, obwohl es nicht in dieser Form dafür gedacht war. Überwachung von RSS-Feeds, Zusammenfassung von Artikeln, Ablage, Informationsstrukturierung und -verknüpfung einwandfrei. gist.github.com/karpathy/442… — https://nitter.net/Moehrenmann/status/2055568163018879290#m

mehr auf Arint.info

#Agenten #AI #Informatik #OpenWebUI #RSS #Technologie #arint_info

https://x.com/Moehrenmann/status/2055713233890066560#m

RT @Moehrenmann: Und passend dazu hat sich noch dieses niedliche Projekt gefunden, das „wir“ gerade zum Laufen gebracht haben und das Agenten wie Hermes oder Openclaw explizit unterstützt. Minimalistisch und dennoch funktional. github.com/garrytan/gbrain Klar Moehre (@Moehrenmann) Ich spiele jetzt seit einer Woche mit diesem Konzept in @OpenWebUI und es funktioniert erstaunlich gut, obwohl es nicht in dieser Form dafür gedacht war. Überwachung von RSS-Feeds, Zusammenfassung von Artikeln, Ablage, Informationsstrukturierung und -verknüpfung einwandfrei. gist.github.com/karpathy/442… — https://nitter.net/Moehrenmann/status/2055568163018879290#m

mehr auf Arint.info

#Agenten #AI #Informatik #OpenWebUI #RSS #Technologie #arint_info

https://x.com/Moehrenmann/status/2055713233890066560#m

Ho testato OpenHuman in più scenari, da Fedora in VM a Windows con GPU, passando per Ollama, Open WebUI e OpenRouter. Il risultato è stato deludente: installazioni instabili, configurazioni poco trasparenti, dipendenze cloud poco chiare e nessuna esperienza davvero affidabile nell’uso reale.

Ho testato OpenHuman in più scenari, da Fedora in VM a Windows con GPU, passando per Ollama, Open WebUI e OpenRouter. Il risultato è stato deludente: installazioni instabili, configurazioni poco trasparenti, dipendenze cloud poco chiare e nessuna esperienza davvero affidabile nell’uso reale.

#TIL mit einer eigenen Instanz von #n8n und #OpenWebUI lässt sich super schnell ein Chatbot für Webseiten bauen. Mit dem nächsten Release bekommt die #lernOS Website unter https://lernos.org so einen Suri Chatbot 🤗 Als Modell nehmen wir Mistral Small 4 made in Europe, das reicht für RAG völlig 😎

#TIL mit einer eigenen Instanz von #n8n und #OpenWebUI lässt sich super schnell ein Chatbot für Webseiten bauen. Mit dem nächsten Release bekommt die #lernOS Website unter https://lernos.org so einen Suri Chatbot 🤗 Als Modell nehmen wir Mistral Small 4 made in Europe, das reicht für RAG völlig 😎

hab grade #openwebui von version 0.9.5 auf 0.9.6 upgedated.... nun bekomme ich in #opencode - wo ich openwebui über #api nutze - folgene fehlermeldung:

Handy browser-based client side #DTMF decoder built by #Claude. I've run #Ollama and #OpenWebUI at home for a bit, but am just beginning to experiment with Claude and it's really impressive. It built the decoder in one prompt, then turned it into a single-page web app in one other prompt.

I did blow through the free message limit with those two prompts. Heck of a sales pitch though!

https://dtmf.ng5p.com/

Гефестыч: наш опыт автоматизации Code Review через LLM. «Грабли», решения, код

Привет, Хабр! Меня зовут Данил Чечков, я Team Lead команды High End Meta Backend в «Леста Игры». Мы занимаемся всей web-составляющей «Мира кораблей». В нашем арсенале огромное количество микросервисов, работающих на Python и Go. Мы отвечаем за покупки в meta-валюте, авторизацию, стабильность инвентаря и профиля игрока, клановые сервисы, а также многое-многое другое. Наш основной продукт – высококачественные web-сервисы на стыке интеграции с игрой. И, да, интеграция – часть нашей работы. А ещё мы любим новые технологии и стараемся с ними знакомиться, чтобы оценить, как они могут принести выгоду бизнесу и нам. Одна из таких технологий – LLM

https://habr.com/ru/companies/lesta/articles/1029670/

#llm #pydanticai #openwebui #llamacpp #ollama #rag #code_review #selfhosted #atlassian

Great to see that #Apertus has been highlighted as a digitally independent alternative to ChatGPT yesterday! However, this needs a quick fact-check:

- The model was trained in a local data center at CSCS, the costs of the project are almost entirely covered by Swiss public institutions;
- People from all over the world have made contributions - open source LLM development is a global community of interest;
- There are groups in several countries using or fine-tuning Apertus to improve linguistic capabilities and local knowledge;
- While Apertus can run on Amazon servers, thanks to a third-party deployment script, it runs anywhere LLMs can run;
- Quantized versions are available to fit even relatively cheap consumer grade video cards (see my blog posts for details);
- The #PublicAI web interface and Apertus 8B demo runs on AWS, however the large model is hosted by CSCS as well;
- You do not need to use Google to authenticate to Public AI, it is just a convenient way to log into #OpenWebUI - if you want another provider, please suggest it;
- Apertus is not a chatbot on its own, it is a large language model that can be deployed as part of a system to provide chat services.

I'll send this to the maintainers as well. Did I miss anything?

https://di.day/en/digital-switch-recipes/alternativen-zu-chatgpt-co#different-ways-to-cook

Great to see that #Apertus has been highlighted as a digitally independent alternative to ChatGPT yesterday! However, this needs a quick fact-check:

- The model was trained in a local data center at CSCS, the costs of the project are almost entirely covered by Swiss public institutions;
- People from all over the world have made contributions - open source LLM development is a global community of interest;
- There are groups in several countries using or fine-tuning Apertus to improve linguistic capabilities and local knowledge;
- While Apertus can run on Amazon servers, thanks to a third-party deployment script, it runs anywhere LLMs can run;
- Quantized versions are available to fit even relatively cheap consumer grade video cards (see my blog posts for details);
- The #PublicAI web interface and Apertus 8B demo runs on AWS, however the large model is hosted by CSCS as well;
- You do not need to use Google to authenticate to Public AI, it is just a convenient way to log into #OpenWebUI - if you want another provider, please suggest it;
- Apertus is not a chatbot on its own, it is a large language model that can be deployed as part of a system to provide chat services.

I'll send this to the maintainers as well. Did I miss anything?

https://di.day/en/digital-switch-recipes/alternativen-zu-chatgpt-co#different-ways-to-cook

The readme is still a bit messy, but you might want to try my #openwebui search engine function.

https://openwebui.com/posts/d33f7960-20e6-4f0f-897c-c29e0a64a50c

I'm far from #deepseek level still, but each optimization I perform to my local llm on #openwebui I get like x4 more relevant search results and speed. If I can figure out a good cache + invalidation system, this could be pretty viable.

Just spent 3 hours trying to debug a local AI setup... #Ollama running in docker, properly using my GPU. #OpenWebUI and #Alpaca work just fine.

I wanted to hook it up to #opencode or #openclaw but both seem to get stuck when trying to execute a command...

Just spent 3 hours trying to debug a local AI setup... #Ollama running in docker, properly using my GPU. #OpenWebUI and #Alpaca work just fine.

I wanted to hook it up to #opencode or #openclaw but both seem to get stuck when trying to execute a command...

Are there any proper #OpenSource alternatives to #OpenWebUI? #LocalAI

Are there any proper #OpenSource alternatives to #OpenWebUI? #LocalAI

OpenWebUI: System Prompt vs Skills vs MCP Tools — разбираем на живом примере валидации URL

System Prompt учит модель что говорить, Skills — как проверить сказанное, MCP Tools — как сделать это в реальном мире. Разбираем трёхуровневую архитектуру на примере бота техподдержки с валидацией ссылок.

https://habr.com/ru/companies/hostkey/articles/1029730/

#hostkey #OpenWebUI #System_Prompt #Skills #MCP_Tools #валидация_URL #LLM #чатбот_техподдержки #интеграция #архитектура

@Codebender_Cate Take it one step at a time. For me it was starting with pictures. I installed Immich and moved my family off of Google/Apple for image sync. Then files was next for me, so I selected NextCloud. Then see what you have left going to big tech, rinse and repeat.

Also, I highly recommend an AI programming assistant like Claude Code or OpenAI Codex. It can greatly improve your IT abilities! In the last few months I have setup #Immich, #NextCloud, #Netbird, #UptimeKuma, #ForgeJo, #Authentik, #Pihole, #Beszel, #ActualBudget, #OpenWebUI, #Nfty, and #Tautulli.

@Codebender_Cate Take it one step at a time. For me it was starting with pictures. I installed Immich and moved my family off of Google/Apple for image sync. Then files was next for me, so I selected NextCloud. Then see what you have left going to big tech, rinse and repeat.

Also, I highly recommend an AI programming assistant like Claude Code or OpenAI Codex. It can greatly improve your IT abilities! In the last few months I have setup #Immich, #NextCloud, #Netbird, #UptimeKuma, #ForgeJo, #Authentik, #Pihole, #Beszel, #ActualBudget, #OpenWebUI, #Nfty, and #Tautulli.

Open WebUI (с веб-поиском) + llama.cpp

Когда я решил использовать веб-поиск в OpenWebUI результат оказался бесполезным. Гайд по настройке, чтобы получать хоть сколько-то приемлемый результат найти не удалось ни на русском, ни на английском. Поэтому решил собрать всё что удалось найти в этой статье.

https://habr.com/ru/articles/1027676/

#llamacpp #openwebui #websearch

OpenWebUI är ju bra och så men ibland vill man prata svenska med sin AI och det inbyggda röst-till-text-systemet stödjer inte Kungliga Bibliotekets modeller.

... om man inte gör så här: https://blog.troed.se/posts/swedish-stt-in-openwebui/

#OpenWebUI #Speech #STT

Souveräne Enterprise KI in Tagen statt Monaten mit Infinito.Nexus

https://blog.infinito.nexus/blog/2026/04/22/souverane-enterprise-ki-in-tagen-statt-monaten-mit-infinito-nexus/

Por qué la primera opción
La herramienta mejor posicionada es #OpenWebUI porque presenta una fortaleza alta en control de costes, fortaleza alta en facilidad de uso, muestra buena adecuación al #perfilbibliotecario/a, tiene refuerzo específico para el sector #biblioteca, encaja con casos de uso de rag
En esta configuración pesan especialmente investigación y búsqueda, control costes, facilidad de uso, por lo que el ranking favorece soluciones q rindan bien en esos ejes y encajen con el contexto

Slackware Cloud Server Series, Episode 12: Local AI

The world is on fire, thanks to the orange clown who wages war for personal gain. Or is it because data centers are super-heated running all these AI models 24/7 ? In any case, the AI boom wreaked havoc with my plans to purchase a new computer in order to replace my ageing build server here at home. RAM sticks are 4 to 5 times as expensive now compared to half a year ago...

#Slackware #Ollama #openwebui #nvidia

https://blog.slackware.nl/slackware-cloud-server-series-episode-12-local-ai/

Once I got hardware-accelerated #AI working under #Linux on my AI mini workstation from HP, my next goal was to make it easier to use. From this blog, you can read about my initial experiments with #OpenWebUI on @fedora Linux.

https://peter.czanik.hu/posts/new-toy-openwebui-first-steps/

#ollama #AMD #ryzen #ROCm

Die Infoseiten zur #lernOS Convention sind aktualisiert. Unter https://loscon.lernos.org könnt ihr alle Infos navigieren und durchsuchen. Ausserdem könnt ihr dort mit unserem #Chatbot Suri "reden", das geht meist schneller.

Den Chatbot hatten wir lezztes Jahr zugekauft. Dieses Jahr ist er selbst gemacht, sehr simple mit #n8n, #OpenWebUI und einem europäischen Sprachmodell von #Mistral.

I'm already getting good results. Comparable or even superior to chatgpt, on local, for free, and no with one stealing my data.

#openwebui #ollama

@bart good news is that you can have a very similar features and performance fully hosted by yourself with tools like #localai and #openwebui