LaurensBER

I'm running an [OpenStick](https://github.com/OpenStick/OpenStick) next to my Kubernetes cluster for fun and there's actually a lot of useful software that runs great with a limited amount of ram.

Some of my favourites are:

[Zeroclaw](https://github.com/zeroclaw-labs/zeroclaw) - 3 MB RSS when idle!

[Beszel](https://github.com/henrygd/beszel) - 10 MB agent + 10 MB server

[Gatus](https://github.com/TwiN/gatus) - 10 MB

[Fusion RSS](https://github.com/0x2E/fusion) - 22 MB

[Adguard Home](https://github.com/AdguardTeam/AdGuardHome/) - 40 MB

[Shiori](https://github.com/go-shiori/shiori) - 10 MB

You won't be able to run all of this on a 128 MB ram VPS but I bet you can run 1-2 services easily!

Is it a cause of concern? Yes. Will you get hacked? Most likely not.

Most consumer routers run ancient stacks with all kinds of weird proprietary vendor patches that are tied to a specific kernel version. Looking at you Qualcomm!

However they generally also drop all traffic by default so from the WAN side they look like a black hole. I guess that's "safe" enough for most consumers.

For fun, take a look at the open PRs in the OpenWRT repos. Those people are doing some amazing work but it's also frustrating to see them struggle with things that _could_ be easy if the vendor made some slightly better choices.

I would strongly prefer something like OpenWRT though. Call me paranoid but after the last few weeks I'm increasingly less comfortable with connecting my devices to a blackbox.

Agreed, some LG TVs even hardcode the DNS these days. A simple cron job that resolves the hostnames and blocks the IP on a firewall level is very hard to bypass though ;)

I would go one step further and recommend a router with a customizable, Linux, firmware (OpenWRT, Tomato, DDWRT, etc).

Your TV is probably just one of the many devices that sends a scary amount of requests to a bunch of shady servers. With a firewall + DNS server that you control and a decent blocklist you should be able to block most requests and you won't be exposed if a new device connects to your network (for whatever reason).

The fact that this was proposed and voted on is ridiculous enough. In this day and age we need _some_ politicians with basic technical knowledge to help guide society in the right direction.

This just points out how incredibly incompetent everyone involved is about technical matters (I will refrain from making comments about their competency in other fields but I'm afraid that this is not a great sign).

It's showing exponential growth but mostly towards the value models (DeepSeek Flash, GLM 5.3 Flash, etc). Which might be an indication that these models are "good enough". At this intelligence level perhaps it's more about cost optimisations.

I cannot speak for everyone else but for us (20 people company) AI spending has definitely peaked and is trending downwards. We're consolidating providers and heavily experimenting with "good enough" models.

The issue was that Stack overflow never was enjoyable to use.

Asking questions was difficult and any difficulty questions I might ask would not get a response anyway.

Answering questions was a gamble, sometimes you would get rewarded, most of the time you would not.

> How does that protect anyone that's suicidal?

It doesn't, this kind of language policing is all about control and power.

A friend of mine worked as consultant with a big bank in London. They would get weekly emails about what kind of "newspeak" is allowed or not.

They seriously banned terms such as an "all hands meeting".

Woke 1 was absolutely crazy.

Given how good illegal alternatives like Stremio are this seems like a very dangerous game to play.

You can squeeze customers quite a bit but once they go back to pirating it's incredible hard to get them to subscribe again.

Would you have examples to share? I'm genuinely curious.

> Mechanically it was solid but I just… didn’t care.

This is a great summary of the entire issue with the game.

I really tried to like it but it didn't inspire any enthusiasm. I didn't really care for the plot, the stakes seemed so low compared to the previous title and none of the races/planets really scratched that explore itch.

Another day where the Pareto frontier is moved up and to the left!

I haven't tried this model yet but Deepseek and the GLM 5.3 series do feel like "enough" for a lot of tasks, it's pretty amazing to see that we're still getting better and cheaper models all the time.

I only wished they would focus more on wall clock time. For a hobby project I do not mind that a cheaper model takes twice as long but at work it's hard to justify using a "cheaper" model if the time per task is so much higher than the frontier models.

The amount of time that it took to complete a benchmark is almost never shown in these kind of releases.

In all fairness, only Xiaomi and Deepseek have a long enough cache time for this to matter. As far as I know, most providers use a 5m cache timeout so unless you're working exactly around midnight your cache is more likely to be nuked by a distraction.

For comparison, close to Opus 5. Far behind Sol 6.1 and Opus 5.5.

Probably a pretty decent model for 90% of the usecases!

I've been doing a lot of hobby work with GLM 5.3 Flash and I can count the amount of times that I need a stronger model on two hands.

I have access to unlimited tokens at works and Opus is great but also often overkill if you know exactly what you need.

If you're already using systemd it's barely any effort to add quadlets. Give them a try!

It's easy to break the cache and there's a number of very valid usecases (mostly context management such as compaction or tool shaking).

Not to mention that it can be very strategic to break context. E.g, after chasing a bug for a while it does make sense to summarise all the approaches that did not work so that you have more context available for newer approaches.

Getting 99% cache hit rates while providing a good user experience is the hard part!

Wait until you realize that deep down everything on a computer is text. We juist build "harnesses" to manipulate in-memory "text" in different ways.

One level deeper is all numbers though.

And eventually it's all 0 and 1s!

So it might be a supply side and not a demand side issue?

A debrid service is a paid hosted cache so I guess it usually lives on wherever they can source cheap storage and disk cache ;) the original source is the torrent.

I have never had any issues with the speed though and I regularly stream 50+ GB files.

Podman + systemd is how I run production workloads at work.

For us it's the perfect balance between simplicity, observability (Beszel) and reliability. Not the right tool if you have dynamic workloads but definitely the right tool for boring and predictable workloads.

At home I run K3S, it's awesome. It also breaks a lot and does add a whole bunch of complexity that I would only recommend if you enjoy tinkering and fixing things.

Please give it a try and let us know ;) judging by the benchmark scores I would not expect too much of a difference.

Main difference seems to be that this one is cheaper/easier to host.

Interesting and I'll definitely give this a try, the client/server and plugin system do look very powerful.

However, I've been using OMP in combination with Tern (the terminal) lately and it's an incredibly smooth operation. It's hard to find a decent terminal on Windows (corporate device) but this is the best I've found so far.

In all fairness, depression and suicide have existed before LLMs.

Safety systems are not optimal but improving everyday and although I doubt we'll ever get a perfect safety system, it seems that we're not very far from a "good" safety system. The days of Google recommending glue on pizza and generating diverse Nazis seem behind us.

Not to mention all the untold good that LLMs have done. Personally I've seen that LLMs are incredibly good at pointing conspiracy nuts and extremists into the right (moderate) direction.

In a polarised world, we definitely need that.

Exactly, we're all software engineers here and even for us it took weeks/months to get to a position where we can really leverage LLMs (think platform, security, observability, etc).

Imagine being in charge of a physical factory, sure you might have a 200 USD Claude subscription but you also have a floor full of physical machines that have, if you're lucky, an undocumented debug interface and a paper manual. Not to mention safety inspections, insurance, etc.

You can't vibe code a factory floor. Not yet, perhaps in 5 - 10 years but there are some very hard, non technical, problems that need to be fixed before we'll see meaningful improvement there.

Software engineers are just the first professions on the frontline. So much "work" is repetitive and things like accounting and compliance are often for 60-80% "tick the box" jobs. Legal is another example that comes to mind, I've had Fable 5.1 review contracts before sending it to our in-house lawyer and the quality is easily 85-90% but with a turnaround time of minutes instead of weeks.

AI will have a massive impact, engineers might be the canary in the coalmine but no doubt other profession will follow soon.

To end in a bright note, software architecture and cyber security are going to be growing massively.

$5 as in 5 dollars? I'm genuinely curious, how did they get to that number? I could understand 20 since that's the price of the cheapest AI subscriptions but 5 seems arbitrary and unusable.

FSR 4/DLSS/Xess is something else than a simple spatial image based scaler though.

If you haven't tried machine learning based upscalers do give it a try! The newest generation (DLSS 4.5/FSR 4.1) are in general preferred by users over a native image since they provide a far superior way of native AA.

AI based frame-gen is also very good. 4k/120 Hz (1080p/60 FPS rendered) is very doable on mid-range cards these days.

4k works great when combined with FSR. On a 9070 GPU the frametime cost of FSR 4 is around 2 msec so you can just render the game at 1080P, upscale to 4k.

I usually play like this, it gives a very sharp image at 120-180W with a barely audible fan.

I can't launch Doom 2016, Brothers or Darksiders 2 on Windows. The screen blanks, if I unplug the HDMI cable and plug it in again I do sometimes get a signal. Ofcourse it works flawless under Linux.

Who would've thought that in 2026 gaming under Linux is actually the better option. Pretty awesome to see it work so well!

In some cases chemicals are added to the water to increase cooling efficiency.

This is a choice, alternative approaches do exist but we should definitely not assume that datacenter water usage is _always_ benign.