Hacker News Reader

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

> And I know there's China, but they just get all their capabilities from distilling Claude, right?

I hope this is a joke. But most of the LLM research and inventions come from China? The best papers are from DeepSeek? You either get the data by stealing from humans or distilling from bigger models?

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

Of course. It is a message to both: drop your prices. Opus gous down to 0.3/0.007/1.2 and we will definitely take another look.

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

Yes. Hi. From our team 3/5 of us use 4.1 to do our daily tasks. For paid work for a company who pays us salary. From people around me I hear a lot of my friends being really happy with it especially for the price.

I don't know man, maybe this is not super serious what I'm doing. Some systems stuff with rust, implementing my own desktop apps with iced, porting old DOS games to Linux...

It is a very good model.

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

Luna is very slow and bad at agentic tasks. DS runs circles around it and there are US providers providing cheaper rates no matter the time of the day.

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

Yeah. I've been enjoying Coralbricks 250-350 tok/s speeds and it is hard to go back to slower models.

Lithos promises even faster speeds if you want to pay more.

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

It is super bad on a bit more complex workflows and starts repeating same errors with the same tool until the cycle breaker hits.

6 is worse than 5.6 here.

But it is amazing on generating a report on content generated by better agentic models such as DeepSeek or GLM, which both do a mediocre/bad job on reports.

pimeyson Why isn't the industry freaking out about DeepSeek 4.1 Flash?

Yes. You have to find the provider with pricing that suits your usage.

I am having 98% my input in cache, so using Coralbricks makes sense due to them giving cache reads for free — you only pay for writes. I spend maybe 5-10 dollars a day and my agents basically work day and night implementing things for me.

If your tasks are write-heavy, find a provider with cheaper output.

If you build a customer-facing app, pay a bit extra for 400+ tok/s e.g. on Lithos.

pimeyson Claude Haiku 5.5

I haven't really tried it that much. I use GLM 5.3 a lot, especially with Coralbricks where the cache reads are free so I can let it think a lot without breaking the bank in the following turns. It's really good for bughunt, planning, and security work.

Maybe I should try the flash...

pimeyson Meta and Microsoft take steps to reduce employee usage of Claude AI

You can also ask your employees to bring their own pirated Adobe Photoshop and avoid paying for the license..

pimeyson Meta and Microsoft take steps to reduce employee usage of Claude AI

Yes. Hundreds of dollars per day is easy to burn with Opus if you pay the API rates.

And now if you get the task done faster with the same quality. And the cost is 80 cents, considering DeepSeek and your own server starts to starts to be relevant...

pimeyson Meta and Microsoft take steps to reduce employee usage of Claude AI

It is called paying per token.

pimeyson Meta and Microsoft take steps to reduce employee usage of Claude AI

I would say the rest of the business world is looking for price and ROI rather than the best in class.

pimeyson Claude Haiku 5.5

Yes. We did the same a few weeks ago. Everybody I talk with in the EU startup ecosystem is either doing the same or considering doing it.

pimeyson Claude Haiku 5.5

Yes, it was weird to see MiMo and DeepSeek missing in the article's comparison...

pimeyson Claude Haiku 5.5

Luna is not really the fastest. You need to use it in high/max to get the good output for what it is good for: summarizing. And that is already close to two minutes per task...

pimeyson Mistral Large 4

> You're absolutely right, I should not use a global mutex.

I am eyeing a huge codebase built with Opus. Did multiple passes of refactoring for things like using traits not just free functions. And using the Rust types like PathBuf from the damn standard library not just strings everywhere.

Oh and it is so hard to use a crate. Let's just vibe an inferior version and push another 5k line PR. Nobody reads them anyway right? Logos schlogos, a lexer is easy to write, right?

Maybe I am just getting old, but I remember a time when we were proud on the code we wrote. Now I read some of the old stuff and get a bit sad for what we lost.

Don't get me wrong DeepSeek also does this. But if the model is 100x more expensive I expect better. Of course if you never read the code it's ok. But it does the same mistakes and cannot design great architecture...

pimeyson Mistral Large 4

https://www.lithosai.com/

https://coralbricks.ai/

With structured tool calls you do want to be lenient on parsing, but that is a very small price to pay when the model is 100x cheaper.

pimeyson Mistral Large 4

It is not super good in long horizon tasks and worse than 5.6 in our evals. It really failed the agentic evals where DeepSeek, Kimi and Opus are the winners.

It is great on creating summaries and content.

pimeyson Mistral Large 4

Yes. V4.1 flash performs really well. I don't know what to say, maybe you don't believe, but my team has been using it mainly for almost a month now for programming. It is as bad and annoying as any of the US sota, but costs pennies. And if you look close enough you find providers that can push it 300-500 tokens per second...

Maybe it is due to us being all very experienced devs. And can steer the model. But my daily routine is just to have 8-9 Zellij tabs open, DeepSeek in omp in each, and grind research and code day and night. Really nice model...

pimeyson Mistral Large 4

They will RL it to be better.

What I'm looking for with this is of it hallucinates less that DeepSeek with its sparse attention, and can be used in better summarization and content generation in multiple languages.

DeepSeek sucks with languages other than English and China.

The price is really good with ML4.

pimeyson Mistral Large 4

How I feel it Opus 5.5 and Astra 6 are both quite disappointing. They still write shit Rust. And they are very slow.

Now my provider serves DeepSeek around 300 tokens a second. The code is shit but I can have a few more rounds of corrections.

DeepSeek was maybe a dollar for the full PR. Opus/Astra over 50 dollars. And double that if you use their fast variants which matches the DeepSeek speeds on certain providers.

Yes, it is so good I rarely test new models anymore. Or think about cost.

pimeyson Pixel 11 doesn't yet meet the GrapheneOS security standards and may be skipped

Yubikeys can store 2fa codes. You just install their app from F-Droid and show the NFC key to get the codes.

Now the codes are with you and not in your phone or in the cloud.

pimeyson Court agrees with EFF: Utah's VPN law demands a technical impossibility

Being the writer of the parent comment who started this discussion, gay porn was the open I was thinking about. Then it the whole discussion was about harming women.

BTW all the women I know in my age group watch porn too...

pimeyson Court agrees with EFF: Utah's VPN law demands a technical impossibility

Why are people so upset about porn? It's probably still the biggest portion of internet traffic. Kind of like human nature, don't you think?

pimeyson GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price

Work for my company. Research, code, analysis.

pimeyson GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price

It's like choosing between vim and helix. I started my career with vim in the early 2000's, customized the whole thing and had my config in a version control.

Then I installed helix and I just use it without config.

If you like configuring things take pi, if not omp is pretty much great defaults.

pimeyson GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price

https://omp.sh/ has amazing defaults and it sips tokens. Works really well with DeepSeek V4.1 Flash.

Use the model through a fast and reliable provider such as Fireworks directly, skip OpenRouter.

pimeyson GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price

You mean 4.1 Flash which is the first great Deepseek? The one that actually surpasses Opus in my books now.

pimeyson GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price

How on earth you can do 100 dollars in 2-3 days with DeepSeek? I have 7 agents in omp running 24/7 every day. I use maybe 10-15 dollars a day. A rarely see a session going over 2 dollars. My maximum is maybe 3.5 dollars and that session took three days.

What harness you are using?

pimeyson [dead]

Wdym? If you have any LLM features in your app, how much you want to charge your customers? 100 euros per month? 1000? Or could you just offer them in the cheapest 10 euro package and don't care about cost that much?

There is no reason to pay API prices for Anthropic and OpenAI if they don't cut them down 100x at least.