Paper, coins and FM radio.

I would love to see Laguna.xs-2 on that, because it's very good on cpu only.

Continue on vscode with Ollama running (start it with "ollama serve") is great. There are some offline models like these that im using but not forget the qwen3.5 coder also.

"ollama list NAME ID SIZE MODIFIED laguna-xs.2:latest ba9ecde43b0e 23 GB 12 hours ago nemotron3:33b f6d8b7ff496c 27 GB 4 days ago qwen3.6:latest 07d35212591f 23 GB 6 weeks ago gemma4:e2b 7fbdbf8f5e45 7.2 GB 7 weeks ago gemma4:e4b c6eb396dbd59 9.6 GB 7 weeks ago "

You can download it from Continue or just use "Ollama pull <name>" from what you choose from ollama.com site and search on models. these run mostly on cpu as my 3080 cannot load those with more than 10gb but the cpu speed is amazing, it outputs faster than I can read!

The new laptop has only 32GB DDR5 and a RTX 4070 with 8 GB GDDR6 on Xubuntu, so Gemini recommended qwen2.5. I don’t think I want to run anything larger because as you said it’d run on CPU and system RAM. As it is, the 14B model will still spill over some and not entirely fit into the GPU.

Good tip "we suggest staying with uBlock Origin" but you still do need a Pi-Hole. It is that critical for a minimal setup to even open a page on any browser. I wonder about what browser the Firefox CEO use. I have a guess.

I don't run a pi-hole, but I do use a public ad blocking DNS server, in addition to using ublock origin.

I only run ublock origin and I don't see ads. How would Pi-Hole help me?

A major use case seems to be ads in phone and tablet apps.

I also use ublock origin in Firefox on my Android phone though!

Makes sense if you're on iphone.

A browser plugin doesn't help if you have apps with integrated ads. (Btw, I know users who configure Android to use a publicly available ad-blocking DNS server instead of Pi-Hole)

Ublock origin is available on ios Safari and you can configure custom dns on ios. Just seems weird you guys only mention android

BTW, LMStudio and a few others are really amazing. They allow you to download models from HF and manage many details before load them. A medium pc with an 8 or 10gb graphics card is already a nice setup to run many models, that are really good. You can also run Ollama that is very simple to use and help you code on vscodium with Continue. Pretty nice!

Local AI does not mean privacy or offline. Claude code does not run offline. It needs an internet connection.

"./claude-2.1.126-linux-x64

Welcome to Claude Code v2.1.126

Unable to connect to Anthropic services

Failed to connect to api.anthropic.com: ECONNREFUSED

Please check your internet connection and network settings.

Note: Claude Code might not be available in your country. Check supported countries at https://anthropic.com/supported-countries"

Let me also add that most of services that are private, will connect to the internet. LMStudio and many others will try to get a connection and all others. I don't remember a single one that does not connect to their servers and send some kind of information.

Nice article. Thank you.

Nice, simple, light and dark theme, a calendar. I would remove the internet checks for your storage.ko-fi.com and it did some calls on pypi.org and files.pythonhosted.org. All it needs should be on the docker by default. But that is just me.

Thanks for the feedback! Appreciated!

----

I had some time to look into this, I think it is partially valid.

The pypi.org and pythonhosted.org calls happen only at docker build time when dependencies are installed, the running container makes no calls there.

The ko-fi badge image (storage.ko-fi.com) is a real runtime outbound request though, every page load fetches it from their CDN.

That's fair criticism and it has an easy fix which is to self-host the image, which I'll do like I do with the rest of things, like fonts.

Thanks again!

I find our standards and norms around network access to be unacceptable. Programs shouldn’t access the Internet without opt in by the user either as a setting or a UI interaction.

We had many years of applications from the dial up error and prior that didn’t assume constant connectivity and we all survived.

Even an OS enforced blend of oauth style initial permission and LittleSnitch where the user is asked what they consent to at first launch would be fine.

Hyundai Kona (renamed in Portugal to Kauai)

Hum, I will use perplexity to find what is the mistery in here. Remember that chromium is trying to open perplexity.ai and I don't know what triggered it, even if it was something outside of chromium. So far, I cannot blame anyone... yet. update: I also cannot exclude Firefox that I use at the same time and opensnitch caught the request but shows chromium instead of firefox.