trilogic

An opensource, simple app for windows with 1 click download for offline maps. We do not like it Complicated :) Available in github and Hugston.com

Enjoy

Sure there are many more, but the post is about Local AI Apps not Proprietary or online services, meaning you can download it in your computer and use it without the need of internet connection, payment or any restriction.

The best Local AI apps worldwide 26 Dec 2025 according to ChatGpt 5.2, using this parameters as comparison: Evaluation criteria:

3-click install → load → run

Install scope (User vs System)

Privacy enforcement (offline switch, no telemetry, no account, CLI)

Workspace features (files/images, code editor, tables→CSV, terminal)

Open model ecosystem (load models from any folder)

Forced updates

Double memory usage

Code preview option

User-activatable local API

Open-source availability

Legend yes / strong partial no drawback

Ranking Rationale (Concise)

HugstonOne (not a simple wrapper) Only app that on top of the other apps does:

have double memory (1 in chat-sessions and tabs and another in persistent file),

installs as user, not in system or admin

enforces offline privacy, with a online/offline switch

supports open models from any folder, not close inapp ecosystem

provides a full agentic workspace (editor, preview, files, tables→CSV, structured output),

exposes a private local API in CLI beside the server.

LM Studio Excellent runner and UX, but closed source, forced updates, and limited workspace depth.

Jan Open source and clean, but workspace features are thin and updates are enforced.

GPT4All Good document/chat workflows; ecosystem and extensibility are more constrained.

KoboldCpp Powerful local tool with strong privacy, but no productivity layer.

AnythingLLM Feature-rich orchestrator, not a runner; requires another engine and double memory.

Open WebUI UI layer only; depends entirely on backend behavior.

Ollama Solid backend with simple UX, but system-level daemon install and no workspace.

llama.cpp (CLI) Best engine, minimal surface area, but zero usability features.

vLLM High-performance server engine; not a desktop local-AI app.

We at Hugston are working hard to make AI useful for everyone. Today we are releasing Quanta, an application for windows that can convert and quantize AI LLM Models in GGUF.

Available at, https://github.com/Mainframework/Quanta or https://hugston.com/uploads/software/Quanta-1.0.0-setup-x64.... (the newer version).

If you like our work you may consider to share it and star it in Github.

Enjoy

Thank you, all the best to you too.

That´s the whole point, you see!

You are not surprised at all, even though you know for certain the company it is already doing it everywhere. So we should assume that this is normal? Isn´t this concerning?

This is unacceptable for me, I will not pretend that is all right, and I am doing something about it.

You are entitled to your opinion and me to my facts.

If the point of discussion here is to downvote me, there you are correct (cause the post I submit keep getting points but my karma goes down, hah funny).

Maybe I am in the wrong place, and so are all the lawsuits that claim otherwise.

Well as you may see this was not just an idea but was implemented before chatgpt. https://www.reddit.com/r/LLMDevs/comments/1ou8rvp/comment/no...

And this is not the first time as mentioned. Absurd is you if you think that the chats don't get logged and filtered for certain purposes.

How do you think they report to Law enforcement certain crimes??? Keep your attitude for you please.

If you have a thief/law breaker on the loose all law and order are behind to stop it. If it harmed an amateur or professional is irrelevant.

What is relevant is this:

That is even more concerning, means that every gov employee should apply same logic, but as we know it is not. Factually speaking, whatever info exists in internet, has been scrapped and used for different purposes, violating the law in plain sight. Therefore we should assume that will happen the same with whatever becomes digital data in the future, and we know how all is towards digitalization (example the currency banks, the social security banks, the IRS etc). Furthermore, how difficult will be to reverse engineer an app to get the source code and get around it with intelligent ways (avoiding patent, design protection). It takes a whole life for a human to reach eureka moment but 1 sec for AI to steal and use it. We need better solutions, because when AI takes live in robot machines, it may be late. I have a possible solution (among others) which is guaranteed to work, (AI sensors which will cause pain for every wrong action, same like we have when put the hand on fire), In my opinion, need to be implemented asap as a standard globally.

Sarcastic behavior detected :)

Ask ChatGPT that, It will stay thinking for hours, not being able to answer.

Try it yourself... If that is a llm issue in general why this small model gets it right first shot? (check link).

Testing all the GGUF versions of Qwen3 VL from 2B-32B : https://hugston.com/uploads/llm_models/mmproj-Qwen3-VL-2B-In... and https://hugston.com/uploads/llm_models/Qwen3-VL-2B-Instruct-...

in HugstonOne Enterprise Edition 1.0.8 (Available here: https://hugston.com/uploads/software/HugstonOne%20Enterprise...

Now they work quite good.

We noticed that every version has a bug:

1- They do not process the AI Images

2 They do not process the Modified Images.

It is quite amazing that now it is possible to run amazing the latest advanced models but, we have however established by throughout testing that the older versions are to a better accuracy and can process AI generated or modified images.

It must be specific version to work well with VL models. We will keep updated the website with all the versions that work error free.

Big thanks to especially Qwen, team and all the teams that contributed to open source/weights for their amazing work (they never stop 24/7, and Ggerganov: https://huggingface.co/ggml-org and all the hardworking team behind llama.cpp.

Also big thanks to Huggingface.co team for their incredible contribution.

Lastly Thank you to the Hugston Team that never gave up and made all this possible.

Enjoy

PS: we are on the way to a bug free error Qwen3 80B GGUF

OpenAi managed to keep Microsoft as major shareholder. Looking at the speed of growth Nvidia and friends are growing, OpenAI is 1 click far to IPO and one more click to 10x after IPO then finally buy Nvidia when GPU market cools down.

Just sayin...

Maybe you are right, on ignoring a study weaknesses and red flags is bad scientific practice. Is that I have been involved personally so long in this topic, that I know for a fact that the results are Good. As I also know many good studies being ridiculed and buried on purpose. No one in the scientific community would dare to criticize a paper in that way. Constructive criticism is connected to intellectual, educated minds, all the rest deserves the same coin or being ignored. I still don´t understand why that comment on top, (I have seen this to many times).

This was meant for Gwerbret (but he deleted the comment). Now is to whom may concern :) Standing by your words you think this paper is shady and you are questioning the work and results behind it. Moreover your comment somehow is on the very top it misleading the users or at least ridiculing the paper. Answering to you: It is indeed very much connected to the LEVELS of Vit D not the absence of it. You fail to understand and acknowledge the importance of the results (even though you already know and confirm the benefits of Vit D). Regulating the levels of it (keeping them higher then average) it prevents health issues by regulating many biological functions/pathways, raising the immunity and lifespan in general. This is the real cure which prevent incredibly terrible future health issues and suffering.

Edit: Just for this effort, this paper deserves Credit. Bravo.

Great job, Thank you.

We support your work and offer backup and distribution. Here a copy just in case: https://hugston.com/uploads/llm_models/EuroLLM-22B-Instruct-...

It was time, congrats. What´s the cap of full memory?

> like jumbled stream-of-consciousness LOL, that was I good one, I agree :)

Why is it so hard to understand... The domain and the website is not ours anymore, we let it go June 2025. Someone else bought it and is using it with malicious intentions to discredit and damage us. The website offers illegal services but in the description is talking about our company. Is hiding behind Cloud Flare with HQ in San Francisco. Is not anymore in our hands, now the authorities are investigating it.

My website never served any of that. You got to stop using this very well known techniques by associating further names (knowing perfectly that HN comments get indexed very fast). The domain in question Sverken.se was always used for science or respectable business. Please refrain mentioning otherwise as I repeat, at any point my company website did such thing. Whoever keep associating them further is doing so by malicious intentions.

Made this one for you: https://vimeo.com/1125804680

My clients (which are most the Gov, Uni, Med, are not happy for it but they got over it. My proposals for funds/contracts suddenly get denied unless I explain what is really happening. This is not only damage to reputation but also to business. I repeat this has to stop. Now the competent Authorities are investigating properly but as always they get off with a fine (which is in Billion btw).The choice for them is Either stop this or pay billions, The world belongs to everyone.

What do you mean the facts are not very hard. If you understand just a bit of LLM you will clearly identify a Distill LLM. All the point is to delete the user because he had created some incredible models beating The best proprietary models out there. The pretext was that he was a fraud and his models were not a distill but a copy. Therefor I show clearly that he was framed and his work deserves credit. I still have the models and scripts for all the procedure. I am not interested to argue further, but it is clearly visible to me. The post in HN also is well voted and most importantly commented, meaning that users have a lot to say about it.

That´s is incorrect, I demonstrated how all that is not true. The models were 100% distill and very powerful. Hard facts

https://vimeo.com/1125658282

https://vimeo.com/1125657672

https://www.reddit.com/r/LocalLLaMA/comments/1o1m9nz/the_mos...

This is has to stop right now as it has gone on long enough. Reddit, Google, ProductHunt, Youtube and friends are continuously using their dirty, unethical even illegal techniques driven by profit. I have experienced all myself and I can confirm that the writer is 100% correct. He forgot to mention though that all this is driven by the same agenda, the same people that want to control the narrative. I Wrote about it too : https://medium.com/@klaudibregu/hugstonone-empowering-users-... Now OpenAI joined the agenda and they are playing dirty very hard also. Yesterday Huggingface Deleted the account of a talented User @BasedBase which was creative in open weights (threatening the big techs). The same boot army discredit and reported his work in Huggingface and Reddit till all his accounts were off. They have done the same to me personally since ever started with Hugston.com and HugstonOne. Just Try to google my Company name "Sverken" (that was associated with Hugston.com) It comes out Porn and prostitution services. Even though this is illegal and screw our reputation Google thinks that this is legit and wont take down the information. Instead they decided to put it in the first page ranking first. I have made some calculations and HugstonOne it is indeed very threatening to big techs. If Our Local AI App takes away only 0.0001% of users from proprietary model websites OpenAI, that is a huge amount of money. And that is just one of them. They have tried everything possible to shut us up, to suppress and undermine our work, to discredit us in abusive ways, but they wont succeed. Thank you for speaking up, hope many other do as well. I really wish you get on your feet soon and the best of luck.

IT was a live feed, is so frustrating I cant find the speech anywhere else.

Live now with Emanuel Macron. Europe is finally waking up, finally a light at the end of the tunnel, I am hopeful that a better future is coming. Macron is actually THE MAN in Europe (thank you).

Viva la France, Viva Europe.

ChatGPT with ads, the beginnings...

We communicate via email exclusively with our already users and clients which in 99% of the cases contact us first (from same gmail address that mark our service as a spam), we don't have time to spam which anyway will damage our reputation. The funny thing is that, All this happens mostly with traffic in our own continent in our own country (we are not even trying to get intercontinental clients)!

The thing about selfhosting email is that, Even after getting rated 10/10 in every mailchecker service and after a nearly 1 year service, Gmail still marks it as spam. So Gmail is basically the monopoly that undermines small startup business by marking their Email service as spam. But ofc this will happen because Europe allows it, making Europe lawmakers an inferior entity even to a private company.

>And how do you watch content on dailymotion and vimeo when the content creators only post them on YouTube? I just don´t watch it. My time is too precious to loose it in ads and nonsense policy.