Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DeepSeek v4.1 Flash

twitter.com

Post reply on HN
Print view View original

391–400 of 405 posts

Re: DeepSeek v4.1 Flash

#391

Post by user43928 »

pixel_popping wrote: post #333
user43928 wrote: post #277

Earlier quoted context omitted.

If it's actually comparable in practice that would be very impressive. I am yet to try a DeepSeek model. From the pricing, it's 3x cheaper on cache, 1/3 more expensive on input, and equal on output compared to GPT 5.6 Luna. I would love to compare these two at work, where I pay API prices. At home I will stick to Astra and Fable.

Just out of curiosity, may I know why you pay API price at work versus using dozens of subscriptions that you APIfy?

I work in enterprise

Re: DeepSeek v4.1 Flash

#392

Post by HanClinto »

howunfortunate wrote: post #380
varispeed wrote: post #350

Earlier quoted context omitted.

In this case "safety" means how to restrict access to good models for working class. You can be sure the rich have access to unrestricted and uncensored models.

I really don't think this is true at all. Do you have any evidence to suggest fully unrestricted frontier models are available for a price? Or...even exist?

Yes, this is well-documented and publicly advertised. In Azure Foundry, the feature to modify (or completely remove) safety guardrails and content filtering is called "Limited Access" [0], and one must submit a form to request permission to use this feature. This is one of the more straightforward paths to get access to unrestricted frontier models, but it's far from the only way.

[0] - https://learn.microsoft.com/en-us/azure/foundry/responsible-...

Re: DeepSeek v4.1 Flash

#393

Post by tiborsaas »

lionkor wrote: post #16

I'm a big fan of DeepSeek. Also, ask it what model it is :) In Pi (pi.dev), it tells me it's definitely Claude by Anthropic, via the API via curl it tells me it's "probably ChatGPT", its very funny.

Sure, we will solve the alignment problem soon, then we can probably teach them "who" they are.

Re: DeepSeek v4.1 Flash

#395

Post by dgacmu »

k9294 wrote: post #286

I'm surprised more people aren't talking about the cache hit price: $0.003 per million tokens. I have a feeling that the price of 1 million tokens transmitted over the internet is more expensive than cache hit. Are we close to making the chat completion API obsolete because the cost of context transfer over network is going to dominate the task total cost? Here's the same token usage priced at different rates: a real…

Bandwidth is really cheap in bulk. You can get a 100 gigabit internet connection for about $10,000 a month. If you were to somehow keep that saturated 24/7, you'd move about 30 petabytes in a month, so your per gigabit cost is only $0.0003.

Realistically, if you had it 5% utilized, those million tokens would cost you about $0.0000062, which is pretty insignificant compared to what they charge you. (Assuming one byte per token, ignoring compression)

Re: DeepSeek v4.1 Flash

#396

Post by gandreani »

pampas wrote: post #161

I've run some evals on my puzzle game https://redactle.net/llm-leaderboard Deepseek v4.1 flash is able to solve it some of the time. I've found it burns through more reasoning tokens than any other model. Google models like Gemini 3.8 Flash are still dominating and is able to one-shot most evals while being the cheapest. I'm curious what other unique evals people are running.

It's so bizarre having a low score be GOOD. It's like reverse intuition. Shouldn't it be called `score error` or something along those lines?

Re: DeepSeek v4.1 Flash

#398

Post by eleventen »

torginus wrote: post #214
1f60c wrote: post #206

Earlier quoted context omitted.

> welfare We should be paying attention to it just in case it ends up mattering enormously. It's cheap insurance. Aside from that, US labs' system cards have been pretty useless for a while—I think the last great one was the combined system card for Claude 4 Sonnet and Opus.

I always talk to models using grugspeak, like 'where getcontext used' I felt a bit bad about it, then I learned yday that model's internal thinking traces are also like this

Same. I use this to fix claudespeak. Any hypothetical token savings are just gravy.

https://github.com/JuliusBrussee/caveman/blob/main/skills/ca...

Re: DeepSeek v4.1 Flash

#400

Post by orbital-decay »

jstummbillig wrote: post #314
kouteiheika wrote: post #13

It's so refreshing to see DeepSeek's tech report[1] full of juicy details; meanwhile, something like Fable's system card[2] is like 70% "safety", 10% "model welfare" to make sure little Claude isn't distressed, and 20% benchmark numbers. [1]: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/... [2]: https://www.anthropic.com/claude-fable-5-1-mythos-5-1-system...

If you are distilling from other models (according to Anthropic reports they are [1]), there are probably a bunch of things that you can just do away with. [1] https://www.anthropic.com/news/detecting-and-preventing-dist...

That's one of the reasons why you should never trust a single word from Anthropic and OpenAI (Sam Altman also blamed them back in the day of R1, in a pretty convenient moment). If you know anything about Claude, DeepSeek, jailbreaking, and distillation, you know the claims are clearly bullshit and the models are nothing alike, and forensic attempts agree, in fact we just had another one [1] [2].

Meanwhile, DeepSeek makes their models and methodology open, so Anthropic can (and likely do) grab without giving back.

[1] https://gist.github.com/wsxiaoys/e0286dc6bb624ff5fdf49e7f4c5...

[2] https://gist.github.com/wsxiaoys/102e8654c14d5d27b7b77532026...

Post reply on HN