HNHacker News
TopNewBestAskShowJobs

geek_at

3,699 karma · joined March 5, 2015

submissionscomments
geek_at··on Claude Haiku 5.5
This is literally for you to get tangled in their api and when they stop giving you the allowance they hope you will just continue to pay
geek_at··on Are Babies Conscious?
Of course
geek_at··on US sanctions force The Netherlands off Microsoft and toward alternative NixOS
Because it was suse and fedora
geek_at··on Litelm: LiteLLM Without the Bloat
And let's not forget that they regularly break things. For example they had merged a pull request that was supposed to fix issues related to the openrouter/free models but in the process broke openrouter for all models except the free ones.

The fix was deployed 2 weeks later (!!) to main (but you could downgrade of course).

Or that other time they broke model selection if you had selected "this key can used all models of their team" than the only model in the auto-selection for harnesses was an invalid "all-team-models" entry. Fixes this one in 1 week though.

all of them on the :latest docker tag btw

geek_at··on Qwen 3.8-Flash-Next releasing tomorrow (125B a6B)
The best solution to this for me is to self host litellm or a different router and use model aliases. For example I have a model called "coding" and when a new good model comes out I just switch the backend without needing to change the alias or the key in my projects (opencode, etc).

I have a few of them even a smart router called "agents" which will use local models but if it thinks the request might require higher reasoning it's routing to a different model

geek_at··on OpenRouter is joining Stripe
and pretty sure they're gonna drop crypto currencies as payment on the platform too
geek_at··on Qwen 3.8 27B is excellent, but it defaults to overthinking things
and memory bandwidth
geek_at··on Qwen 3.8 27B
Do you mean 3.6 27b? Because qwen 3.7 didn't have an open weight version
geek_at··on Google fixed more Chrome bugs in June than over the past two years, thanks to AI
won't forgive them either for making it harder to filter out ads
geek_at··on DeepSeek-V4-Flash Update
it does! I'm always amazed when I use DSV4 flash for coding or server checks and after an hour of working with it my (pure api call) balance is about 30 cents
geek_at··on Self-host your mail server
Sorry the title should have been "You should self-host your mail server" but it seems HN removes the "You should" in the beginning of a title.

I wanted to start this discussion because I think we are prime for more self hosted mail servers in 2026 and in the post I explain how to use a locally hosted LLM to nudge the spam detector rspamd to whether or not an incoming mail is spam.

Also in times where the EU can order your mail providers to scan for unwanted content [1] (currently limited to CSAM, later probably more) it does feel better to have your data at your hand.

[1] https://en.philenews.com/international/privacy-to-end-in-the...

geek_at··on Exploit brokers pay $500k for WordPress RCEs. I found one with GPT5.6 and $25
it would help if they used strict types and modern standards but as you say the wordpress codebase is beyond dated and held together with duckt tape
geek_at··on Telegram Serverless
I use a signal api on a daily basis for all my agentic chatting and notifications about energy prices and server downtime notification.

Setup was quick enough but it needs a real phone number once during setup which is kind of a hassle. still better than using telegram which has no encryption for the bot backend so the devs can read everything you write

geek_at··on Podman v6.0.0
don't forget `alias docker='podman'` for muscle memory
geek_at··on Opening up 'Zero-Knowledge Proof' technology to promote privacy in age assurance
Of course not? The idea would be a government (who already has your age data for example) will allow you to create a signed message and the platform you are verifying your age to doesn't know who you are, what age you are but that you are of age
geek_at··on Liquid AI reveals 8B-A1B MoE trained on 38T
good point I have the feeling larger models (20b+) rely too much about their stored knowledge and sometimes fail to use tools because they think they know the answer. smaller specialized tool calling models could be the smart route for the future
geek_at··on Qwen3.7-Max Ran for 35 Hours on Unknown Hardware and Achieved a 10× Speedup
https://xcancel.com/mohitgeryani/with_replies
geek_at··on Time to talk about my writerdeck
honestly this explains so much about many blog posts or journalistic articles I read in the last year
geek_at··on Elon Musk has lost his lawsuit against Sam Altman and OpenAI
why would he run Anthropic?
geek_at··on W – The European social network for verified humans
Unless you're not blind to all things we have because of them
geek_at··on Leak reveals Google's Aluminium OS with a 16-minute video
The fact that this os might teach Americans how to spell Aluminum correctly is kind of hilarious to me. Or are they going to mispronounce the OS too?
geek_at··on They Live (1988) inspired Adblocker
Seems not to be available in europe "The uploader has not made this video available in your country"
geek_at··on Warp is now open-source
I was confused by the name. For a second I got excited that OS/2 Warp was being open sourced. That os was ahead of its time. And I don't just say that because my dad was called "Mr. OS/2" because he was very good at selling it to clients of IBM
geek_at··on European AI. A playbook to own it
True we pay for it so we don't have tu subject our own people to the horrible working culture of the US. No maternity leave is diabolical. No money when you're sick for a month? How do you accept this
geek_at··on Personal Encyclopedias
physical books are a greate idea.

I do something similar but with email and more pro-active [1]. I have created my son an email address when he was born and I'm sending him things from our lives and ask family members to to the same. Just to write them about themselves and send photos of their current homes and gardens and partners.

I imagining him looking through his email when he's 18 and reading personalize messages sent by family members who might no longer be with us then.

[1] https://blog.haschek.at/2024/leaving-a-digital-legacy.html

geek_at··on iPhone 17 Pro Demonstrated Running a 400B LLM
Still dreaming about an android keyboard that plugs into local or self hosted llm backend for smarter text predictions
geek_at··on The dead Internet is not a theory anymore
Would that work though? Unless it checks your pulse every 30 minutes I don't see how that would make it better. Bots would use stolen IDs for that. It would only contain it at a smaller scale probably
geek_at··on The dead Internet is not a theory anymore
reminds me of Bill Gates in the 90s when asked about email spam. He said it would make sense to make an email cost like 1 cent so the spammers can't spam as much but this didn't sit right with the mindset of the people at the time.
geek_at··on The Browser Becomes Your WordPress
Interesting! The first use that would come to my mind is using it with a static site plugin and basically using this as a offline and damn simple static site generator or even blogging Plattform.

Looking forward to what others come up with

geek_at··on France Aiming to Replace Zoom, Google Meet, Microsoft Teams, etc.
> the will to absorb the inevitable cost of change

it's simple economics. When US services have to increase their pricese because of trumps tarrifs and these increases are higher than the cost of change, they'll do it. we're almost there

Page 1 of 15Next →