Tepix

First the chat agents, now the kids!

For starters, it’s twice as fast

As is mentioned in the article (now).

That reminds me of a day almost 30 years ago where I spent my last money and had a big pizza gyros. I went from starving to completely satiated. It was absolutely glorius. I’m still a big fan of pizza Gyros thanks to this special memorable occasion.

Food matters.

Sounds nice. I'd like to see a hardware that is optimized for being a low power, high bandwidth network server at home:

- 1 or 2 M.2 NVMe for fast storage

- 2 RAM sockets for expandability. Not sure if DDR5 or if DDR4 is sufficient. SO-DIMM is fine.

- 2.5 GBit ethernet (or better). Two ports would be better than one.

- a CPU with a bunch of cores that can saturate it well. Also a NPU inside the CPU.

- power via USB-PD

- optionally a couple of USB ports for things like USB sticks, Wifi (if needed), printer.

That's it. I don't need a display port like HDMI or DP.

1. In real life, most of us use token packages like OpenCode Go etc.

It would be handy to have a site like this one that takes into account the various deals and attempts to calculate the number of tokens per monthly fee for a chosen model. I realize this makes the task a lot more difficult.

2. It would be handy to have a chart like that for the AI hardware that people own. It helps you decide which model to run (resulting in different levels of intelligence and speed). Also difficult to please everyone (preprocessing vs token generation for example) and to keep updated!

I found https://llm-list.com/ yesterday and when I had a detailed look, I quickly found outdated entries, for example looking at GLM 5.3 flash it listed several providers as "free" that weren't free any longer.

What's absent from this report is the cost. From reading it, I don't think it was cheap.

That’s the bootloader of the original Quest

One solution is likely the same as the one that stops every other iOS update from re-enabling iCloud password sync: Create a profile using Apple configurator <https://apps.apple.com/de/app/id1037126344> that forbids all these AI features and upload it to your devices.

The other solution is to say goodbye to these products that don't respect you.

By the way, if you're annoyed by the new huge "sync" menu entry on Firefox, you can disable it under about:config by setting identity.fxaccounts.enabled to false. Again, a lack of respect, this time by Firefox, for forcing this stuff down our throats. I'm only using Firefox because the alternatives are worse in terms of privacy. Ladybird, we're waiting for you!

I'm not sure yet, tbh. Perhaps it does make sense to outlaw them eventually.

Then again, it will probably not stop someone who is determined. Same as with other legislation really.

The user friendly option would be to print the amount of RAM in very large letters on the PCB. There happens to be a lot of suitable surface on the PCB to the right of the text "Raspberry Pi 5" on the PCB anyway (§). Right now the amount of RAM is hard to read and easy to modify (move a little tiny SMD resistor).

Drawback: You need different silkscreens for different amounts of RAM.

Btw, if you're worried about RAM-swapped units you can already verify them as described at https://geekworm.com/blogs/news/prevent-ram-swapped-raspberr...

--

(§) as can be seen here https://pip-assets.raspberrypi.com/categories/1129-pcn/docum...

Keep a close eye on abliterated and "heretic" open weight models. They will be outlawed first.

I thought you were doing a PR to add an 11th space.

600b-a27b doesn’t sound enticing. Also with the higher number of active parameters compared to GLM 5.3 flash and DeepSeek V4/4.1 flash, I don’t see how they want to be more efficient at inference.

I believe with tailscale you don’t have to trust a 3rd party with your cleartext traffic

They see all the traffic in cleartext. Plus you have to trust them not to maliciously alter your traffic. As a US company, their options may be limited if they are coerced by their government to do so.

Suggestion: Every time someone uses nitter, the post in question is copied over to Mastodon and subsequent accesses just redirect to the copied post. That way you can interact with the copy on Mastodon.

Was just looking over the Safari 27 release notes¹, which is part of macOS 27, and noticed this:

Web Driver

New Features

- Allow your agent to connect to a Safari browser for development and debugging via the Safari MCP server. (176038457)

Related (From July 1st)

Introducing the Safari MCP server for web developers

https://webkit.org/blog/18136/introducing-the-safari-mcp-ser...

So that's interesting. On the other hand, looks like WebXR support for Safari will never come.

---

¹ https://developer.apple.com/documentation/safari-release-not...

Time to dig up 386bsd

Sounds interesting. Clever name, too. And open source. I’ll check it out.

Remember that story a week ago „everything I own, owned“? That’s one way.

You need to search deeper!

Looking at the iPod 6G QEMU emulation story:

> "With the help of Claude Code, I was able to create code to extract it using emCORE."

You're absolutely right.

Make it serverless.

So apparently the neki router uses many parts of Postgres. But is it open source itself?

Apparently not.

I'm sure that it will work, but tokens/s will suffer quite a bit. You may still be able to get 10t/s or so..

Look closely at the floor. You're not moving backwards.

> "the Speed of Light is pretty slow" > How i got there: > The closest major galaxy to the Milky Way is Andromeda, and is 2.5 million! light-years away.

The speed of light is high. It's just that the universe has expanded into an unimaginably large area by now.

Sounds like the same issue with Apple re-enabling iTunes sync of passwords. The only way to prevent it from doing that is to create a profile that prohibits it and sync it onto the iDevice.

Horrible.

You can get cheap wireless earbuds with ANC for 10€.

Apples cost 130€.

How much of that difference is in the BoM? Probably not that much.

Granted, Apple does a lot of expensive R&D. But they also sell a ton of Airpods, so the R&D cost per unit is small.

Apple is making a lot of money on Airpods.