l5870uoo9y

> high energy consumption

Just to address this separately: there is effectively energy rationing in several European countries, where the government intervenes to prioritize who should receive energy first (e.g., the Netherlands and Denmark).

Given that the solution took a somewhat “unusual” approach, I find it even more unlikely that an AI model would have come up with this on its own.

I run a small SaaS[1], like so many others, that uses AI to generate and optimize SQL. Getting this to perform optimally has been a lot of work and now I wonder if OpenAI is outright stealing this knowledge, which without a doubt is highly valuable to them.

[1]: https://www.sqlai.ai

The weekly newsletter is a great way to get the most important information: https://hackernewsletter.com/

Found existing React web frameworks either bloated, buggy or too complex, so I built: https://www.rshono.com/

Location: Copenhagen, Denmark

Remote: Yes

Willing to relocate: No

Technologies: TypeScript, React, Next/Astro/Tanstack, Postgres/MySQL, devops, platform (CF/AWS/Linux), AI

Résumé/CV: Tech lead and full-stack developer with 15+ years of experience planning, architecting, and building web and mobile products. I have spent most of my career inside international startups and scale-ups in Berlin and Copenhagen — leading teams, owning architecture decisions, and helping early-stage companies grow into successful businesses. Please see: https://www.lassetange.com

Email: contact@lassetange.com

Is DeepSeek really behaving different than other Chinese companies? Intellectual theft is ongoing and has been ongoing for decades. Besides security risks and foul play, it is impressive by just how much DeepSeek undercuts OpenAI and Claude. DeepSeek charges $0.87 per million output tokens compared to $50 for Fable and $30 for GPT-5.5.

Not a single word on where the toxic products were produced except third countries?

I wonder how much revenue Stripe would loose if it cracked down on chargeback fraud and free trial abuse (cards defaulting) and similar? Even just a few percentage out of an annual revenue of $5.1 billion is substantial.

> Bia hơi (pronounced “bee-ah hoy” and meaning “fresh beer”) is brewed without preservatives or added carbonation.

Tank beer (tankova) from Urquell is same but it last a week or two in the tank to my knowledge and not just 24 hours as Bia hơi. It is properly the best pilsner in the world.

> The CTO publicly apologised and said capacity needs to scale 30x to keep up with AI-driven load.

I hope they don't start charging for regular use of GitHub, but when I see how some of the vibe coders make thousands of commits a day, I'm becoming more and more skeptical. Would be a real shame if we can't share and cooperate on code for free.

> It's about spending your time where it truly matters.

I think the author conflates productivity with meaningfulness.

> Claude on Amazon Bedrock keeps AWS as the data processor and operates within the AWS boundary. This is a good fit for companies that have strict regional data residency requirements or need their data processed exclusively within AWS's infrastructure.

Seems like there are two different options.

> DeepSeek V4 Pro scores 96.4% on LiveCodeBench and costs $0.87/M output tokens.

Yes and this is a temporary discount which increases to 3.48 USD on 2026/05/31 15:59 UTC.

Source: https://api-docs.deepseek.com/quick_start/pricing

Took a war to realize this.

Kinda feels like a code smell having to optimize your code this way. Normally, data state would be abstracted away in a data store (e.g., Redux), and then specific sub-selections are pulled out using pure functions, e.g., `getActiveUserNames`, `getActiveUsers`, `getTop5ActiveUsers`.

Pho is delicious, but Thai boat noodles soup (guay tiew ruea) is godly; dark, robust, and full of diverse flavors and nuances.

> It could be either a fat-finger mistake when editing the Excel file or deliberate tampering to cover up real data that didn't tell the right story.

I can easily imagine after spending years or decades devoted to discovering a scientific breakthrough that some could be tempted to slightly alter the data. I believe there was some scandal about this a few years back with climate data. Fixing this is however something that AI would do fairly well.

Could be that Claude has particular controversial opinions on eating disorders.

I started freelancing over 10 years ago, and I got my first freelance project through an acquaintance from my former dorm in Copenhagen. After that I got several different freelance jobs through recruiters on LinkedIn. But it was a different time, as the cliché goes. One thing to know about highly paid freelance positions is that they are extremely cyclical. These are positions that mainly exist because the market is booming and screaming for labor, and when that's not the case, companies will much, much prefer to have permanent employees.

In the time before COVID and up until its end, the tech market was booming, but when the economic stimulus ended and interest rates rose, the tech market shrank and the freelance market in Berlin/Germany (and probably Denmark too) has never really recovered. The positions simply aren't there, only very few that many fight over. The great thing about being a freelancer before was that you were almost treated like a rock star, recruiters contacted you all the time, and you could pick and choose.

If you don't have a large network, you can try signing up with different recruiters and see what they have to offer. You might be lucky if you live in regions with less competition and where clients are looking for someone local to work on-site. Getting hired these days also requires that you have (substantial) experience and can show some projects.

(By the way, I’m looking for a permanent position—preferably in Copenhagen—as the bank requires this to approve my loan application. Here is my resume just in case: https://lasse.sometechblog.com/)

My impression the reverse is true when upgrading to GPT-5.4 from GPT-5; it uses fewer tokens(?).

Most VPS hosting companies have cheap entrance VPS but then steep pricing. 8 vCPUs on DigitalOcean cost 96$ and ~11$ on Hetzner.

Has Hetzner discontinued their US data center?

In hindsight, it is painfully clear that Antropic’s conservative investment strategy has them struggling with keeping up with demand and caused their profit margin to shrink significantly as last buyer of compute.

It’s fascinating to read but I have a hard time imagining a public western railway provider could evolve into a train based mega corporation doing real estate and health services.

The absurdity of the climate debate is that “we” talk almost constantly about two energy sources (wind and solar) that in no way have the potential to provide the stable baseload power required to electrify society. And unless nature has blessed your country with abundant geothermal or hydroelectric power, that leaves you with the following options: oil, coal, or nuclear power.

I migrated the landing pages for my app[1] from Nextjs to Astrojs mainly because I was paying Vercel $20 per month for serving static pages(it’s 4 times more than I pay Railway for the Postgres database for the actual app and also 4 times more than I pay Cloudflare for hosting all my apps). I used AI for migrating and it took a few days only as the existing repo was used as “instructions” and it included some upgrades and improvements here and there.

[1]: https://www.sqlai.ai/

Reminded me of "sitting at a dock of the bay" with tamer lyrics and more bland and monotone. I suspect AI is much better at DJing than actual composing.

I can't find any of his songs on Top 100[1], is it another list?

[1]: https://music.apple.com/us/playlist/top-100-usa/pl.606afcbb7...

> Our method, simple self-distillation (SSD), is embarrassingly simple: sample solutions from the base model with specified temperature and truncation, then fine-tune on those raw, unverified samples via standard cross-entropy loss.

So you prompt the base model for answer and then rerun the prompt with the answer from the first run?