Hacker News: nik736

New comment by nik736 in "There are now 5 inference providers for abliterated DeepSeek"

nik736 — Tue, 27 Oct 2026 06:38:24 +0000

Morph is hosting this as well

New comment by nik736 in "Mistral Large 4"

nik736 — Wed, 21 Oct 2026 21:13:43 +0000

Havn't seen it for months with other frontier models.

New comment by nik736 in "Mistral Large 4"

nik736 — Wed, 21 Oct 2026 21:00:23 +0000

Sorry it was not one time but it happened easily 5-10x during the past few days on different topics and prompts and I didn't store them. It felt like when pushing the model on some specifics in some niche topics in German (I have been prompting in German and the model responded in German). It was the acutal output and not the thinking traces and it was through the API. I jokingly asked it what the character meant and it didn't even make sense in the context so it seemed like random chinese characters.

New comment by nik736 in "Mistral Large 4"

nik736 — Wed, 21 Oct 2026 20:06:56 +0000

The weird thing about this model is that it spits out Chinese characters just like the chinese models when chatting with it in German. Happened to me quite often during the past few days.

New comment by nik736 in "Germany's 17M Immigrants: Where They Come From"

nik736 — Sun, 18 Oct 2026 13:20:21 +0000

There were a lot of people born in Russia with German ancestors and they had to leave Russia during WW2 or later left by themselves when the Soviet Union collapsed. Many chose to go back to Germany. This is also why there are probably a lot of Kazakhs in those statistics.

https://en.wikipedia.org/wiki/Russia_Germans

New comment by nik736 in "DeepSeek-v4.1-Flash-1017"

nik736 — Sat, 17 Oct 2026 09:15:43 +0000

Those are crazy numbers for a Flash model. Hopefully they can scale what they are doing to 4.1 Pro.

New comment by nik736 in "Postgres isn't slow, your storage is"

nik736 — Mon, 12 Oct 2026 09:10:49 +0000

When the host dies with local storage, the VM is dead. With EBS it's not, since it simply can be booted up on another host. EBS is also replicated accross multiple hosts.

New comment by nik736 in "Verda (Finland) raises $189M in Series B"

nik736 — Tue, 22 Sep 2026 11:00:58 +0000

Verda is great! The only thing bad about Verda is that it's sold out fairly often :-) So the $189M will hopefully help with that.

New comment by nik736 in "We must pace the frontier"

nik736 — Sat, 12 Sep 2026 15:33:39 +0000

Anthropic trained their LLMs with copyrighted stuff but distillation is bad. Anthropic has closed models, China releases as open weight. DeepSeek even allows distilling their models, but China = bad. Understood.

New comment by nik736 in "Mistral raises €3B"

nik736 — Tue, 08 Sep 2026 15:20:19 +0000

Mistral Small 4 is also a MoE model, with way more params, so I would expect it to perform better. Our benchmark involves around 10% communication and writing in German and English and Gemma 4 26B A4B beats it although creative writing is the only category in our benchmark which might be a bit subjective.

New comment by nik736 in "Mistral raises €3B"

nik736 — Tue, 08 Sep 2026 07:33:09 +0000

Mistral has solid OCR, STT and TTS models and I would love to support them by switching with all of our business workloads to Mistral... but their LLM models are sadly not competitive at all. In our business benchmarks their Mistral Medium 3.5 with reasoning is worse than Gemma 4 31B and Glimmer 30B. It's a 128B dense model that's priced accordingly! Mistral Small 4 is way worse than Gemma 4 26B A4B. I applaude the effort that they release those models as open weight but Gemma 4 models are currently way easier to run with more tok/s and less hardware. Their API pricing is just insane for what you get. But I guess enterprise customers don't care about it, this is why they are probably not lowering it.

New comment by nik736 in "Hy3"

nik736 — Fri, 10 Jul 2026 09:12:21 +0000

Gemma 4 31B is underrated. It surprises me a lot.

New comment by nik736 in "Mistral OCR 4"

nik736 — Tue, 23 Jun 2026 15:30:06 +0000

I have put together an internal benchmark on 1000s of business documents with weird tables, structure, etc. that I run on every relevant model release. Opus 4.8 performs very very well. But it is obviously overkill for the task (and expensive at doing so). I just wanted to respond to the OP.

New comment by nik736 in "Mistral OCR 4"

nik736 — Tue, 23 Jun 2026 15:15:39 +0000

Opus is very good at OCR. Way better than the small 1-4B VLMs. If Opus failed, most likely those smaller models will fail as well.

New comment by nik736 in "Claude Opus 4.8"

nik736 — Thu, 28 May 2026 18:23:15 +0000

Why would you need the 3rd run if you pick the "one in the middle"?

New comment by nik736 in "AI (2014)"

nik736 — Fri, 20 Mar 2026 11:29:03 +0000

> (I originally was going to say a computer that plays chess, but computers play chess with no intuition or instinct--they just search a gigantic solution space very quickly.)

Isn't that how LLM models are trained right now? Trying to predict the next word within a "gigantic solution space". Interesting.

New comment by nik736 in "Winstwaker – automated bookkeeping with a real accountant attached"

nik736 — Tue, 17 Mar 2026 09:50:27 +0000

In Germany we have several accounting software solutions like that for 5-10+ years that integrate with bank accounts, paypal, etc. - automatically suggests booking accounts and exports it via a REST API to the software accountants use. Accountants have access to it. Is this basically the same as your solution?

New comment by nik736 in "Show HN: Better Hub – A better GitHub experience"

nik736 — Thu, 26 Feb 2026 10:51:25 +0000

No lightmode?

New comment by nik736 in "Deutsche Telekom is throttling the internet"

nik736 — Sun, 25 Jan 2026 11:41:42 +0000

Well, we have to "register" every new IP or new mail server with them as well. It's annoying and a weird system, but they respond quickly and it's just one todo we have to think about.

New comment by nik736 in "Apple's slow AI pace becomes a strength as market grows weary of spending"

nik736 — Tue, 09 Dec 2025 15:45:48 +0000

> Meaning that the technology was there and ready to make an experience that was truly excellent

In general I would agree, but Siri is honestly still so bad.