dust42

> We are making this release as fast as possible (we learned of the problem yesterday) because we have recently observed multiple cases of rapid bug rediscovery by AI agents, sometimes within hours, and so now consider long embargo periods to be a dangerous practice that primarily benefits attackers.

I think this will likely become standard. Nevertheless, a few stormy months are ahead.

> being reported and discussed publicly

And yet the article does not reference any of them. So of the 1000 they found, 400 were already public. And the other 600 we have to believe.

Move fast and break things. 4-eyes principle is for brick-and-mortar banks stuck in the prehistoric past of paper money. /s

So Hinton doesn't trust the tech leaders but then the consequence of having governments censor LLMs is that only tech leaders and governments have unlimited access to AI.

If someone manages with the help of AI to hack into a vital system, then the system was already in a critically bad state beforehand. If developers can't use AI to check the code for security flaws, then security gets worse and not better.

It is perfectly well engineered content for the target audience ahead of the IPO. Gotta do something to push your stock options...

The funny part is that you can do such a youtube clone with something outdated like Qwen 3.5 35B, no need for the latest SOTA frontier trillion weight model.

I'd assume that there are many more engineers from the valley on HN thus the votes are accordingly. Look at the sibling comment from danpalmer - he is a Google engineer and thus he has an interest of pushing Google products and trashing open weight models. HY-World is a real 3D model and plays in a totally different league, not Genie 4.

> It is amazing how far you can get by taking credit for accomplishments you didn't do by talking those who do accomplish things into not caring about the credit. --$clever

Feel free to replace $clever with any name of your convenience.

The problem starts, when places like FOSDEM which originally where slightly left-leaning-anarchist (in a positive sense) start talking about censorship but only mention Russia, China and Iran. Maybe you should google for Jaques Baud, who is actually trapped in Brussels and invite him? Yes, yes, I know, Chinese LLMs and Tianamen square. Well, go and ask chatGPT for the crimes of presidents of the nothern area between Atlantic and Pacific... The problem is the global brainwash.

The CPU is for mobile phones. An Nvidia H200 has 4.8TB/s. An RTX 5090 has 1.8TB/s. Both use ~700W - not comparable with a phone.

SEO was yesterday, now it is AIO - you will have to wait longer for the results and likely you will need a lot more cash: pay TIME for the ads, wait until next model release and see what sticks, then rinse and repeat.

As a user I still prefer .pi right there in my home directory.

In 2 years from now we will be at 400%. https://xkcd.com/605/ Also, it is called kernels (you have nitpick in your username)

> As with many conflicts this one goes back a long way. Ukraine gave up its nukes to Russia in exchange for a non aggression pact that the invasion broke.

From the Guardian (12 days ago): "Kyiv’s decision to honour second world war fighters who killed about 100,000 Poles has revived simmering tensions". Bandera was a Nazi collaborator and today's Ukraine politics worships Bandera. Lots of streets are renamed to honor him. Eastern Ukrainians are considered second class - or as some politicians said: "There are 2 million people too many".

The US (look up Nuland and her quote about 'fuck the EU') fueled with $5B the nationalism in Ukraine. Now imagine, the US would fuel with $50B the nationalism in Belgium and the the flemish would attack wallonie. You know what would happen? France would support the wallonie. Nothing to see here.

With a M5 16c 48GB and Qwen 3.6 35B Q4 I get up to 1900 PP/s and 80 TG/s. With an Nvidia 5090 I get 7800 PP/s and 280 TG/s.

Together with pi mono I wouldn't want to go back to Claude & Co. Speed, quality of the answers, short answer times at any time of day - once you have eaten from the fruit your definition of SOTA will change...

For reference, I do software development since 30 years, I am not vibe coding the umpteenth todo list.

This and we don't know yet what happened. It could have structurally collapsed - very unlikely, it could have uncommanded retracted, or maintenance has overridden the protections. I'd place my bets on #3, handling error in maintenance mode.

From the picture and the text this aircraft was parked at the gate. During a hard landing the nose gear may collapse but not while being parked. And while parked there are protections to prevent retraction. However, these can be overridden by maintenance.

I just spent the last two weeks digging into workflow state engines and temporal was one of the candidates. It is a VC backed fork of Cadence. The got 0.3B funding and whatever positive I read about them on the net I take with a big spoon of salt. Just my 2 cents.

For many models the performance of llama.cpp on Mac is 20-40% lower than MLX. Did you try MLX? At least on HF there are MLX 2-bit quants. Unfortunately I have only 64GB, so I can't test it.

The output of any LLM is always 100% hallucination by principle. On top of that, most benchmarks are at best an approximation of LLM quality. Your use case decides which one to use. That said, I haven't tested v4 yet but the old 3.2 is still a decent model. And concerning use cases, I had coding problems that Opus couldn't solve but a local 35B model did.

All the talk about frontier and SOTA is do dig deeper and deeper into the pockets of VCs and finally do an IPO.

That website is so low effort that 2s is actually long to figure it out. Very sure that it is robot upvoted.

Edit: I live in the cheese triangle, France - Switzerland - Italy.

Tinder for indie (hardware) devs and their customers. I.e. a webshop for indie devs who sell small series of niche hardware.

Ask on mikrocontroller.net. There will definitely be people who know who in Germany is doing it or maybe even offer to do it for you. One of the last old fashioned forums on the net. German speaking but nowadays that shouldn't be a problem.

I am using it with pi agent and I have stopped renting tokens. Much better for me than Claude Code, on M1 Max 64GB. This model with oMLX is at 16k context PP 919.9 tok/s and TG 54.7 tok/s. You have to manage the context but the better you manage context the more focused the output is. I use it without thinking.

I once worked for a company that bought a spot in the evening news (french TF1). It worked that way: a french minister was visiting a fair and coming to stop in front of the booth and getting a product demo. And that ended up in the evening news. Since that time I kind of watch the news with different eyes.

The law is still there though and thus its application can be resumed at any time without need of parliament approval.

There is a project on github named ANEMLL. Was discussed here a month ago, running LLMs on iPhone - https://news.ycombinator.com/item?id=47490070

Well, at least we know by now that Mythos is a mythos.

I really love it. The simplicity is key. The first play project I made with it was a public transport map with GTFS data - click on a stop and get the routes and the timetables for the stop and the surrounding ones. I used Qwen3.5-35B on Mac M1 Max with oMLX. It wrote 98% of the code with very little interaction from me. And very useful is the /tree feature to go back in history when the model is on a wrong track or my instructions where not good enough. I usually work in a two path approach: first let the model explore what it needs to fulfill the task and write it into CONTEXT.md (or any other name to your liking). Then restart the session with the CONTEXT.md. That way you are always nicely operating in 5-15k context, i.e. all is very fast. Create an account for pi (or docker) and make sure it can't walk into other directories - it has bash access. Add the browser-tools to the skills and load them when useful: https://github.com/badlogic/pi-skills

No need for database MCP, I use postgres and tell it to use psql.

Occasionally I use prettier to remove indentation - the LLM makes a lot less edit errors that way. Just add the indent back before you commit. Or tell pi to do it.

Context: Mario Zechner is the creator of the pi coding harness which powers OpenClaw. OpenClaw is made by Peter Steinberger, a friend of Mario Zechner. Armin is another friend who made public that OpenClaw is based on pi.

Pi itself is a minimalist coding harness with a tiny 1500 token system message and only read, edit and bash as tools. I only discovered it a few weeks ago and it is surprisingly powerful with a local Qwen3.5-35B - especially as it allows to keep the context low.

Mario's blog posts are not easily digestible (imho) until you have read a few of them but they have plenty of profound thinking. His blog is for me the first one in years where I have spent an hour to read several posts.

Mario is deeply rooted in the OSS system and basically that is what he is talking about here in this post. That said, I have no idea what earendil is doing, except that it is based on pi.

Edit: My personal take - "I've sold out" is very much Austrian style because actually it is the opposite. To quote one thing from the post:

"Then Miguel and Nat approached us. Long story short: we sold RoboVM to Xamarin. A short while later Xamarin closed-sourced our open-source RoboVM core, quickly followed by Xamarin selling to Microsoft. Then Microsoft shut down RoboVM immediately.

While there was some monetary gain, everything about this fucking sucked."

So Mario did a lot of vetting to hopefully avoid this from happening again.

The social media team.