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

US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

reuters.com

Post reply on HN
Print view View original

561–570 of 633 posts

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#561

Post by Bnjoroge »

atty wrote: post #398
Cakez0r wrote: post #395

Earlier quoted context omitted.

Mimo 2.5 pro is the best intelligence / dollar

In most cases it seems (at least to me and colleagues) to be turning out that picking best intelligence is a better option than picking better intelligence / dollar, assuming you can afford the cost. At least on interesting problems. If you’re doing generic web dev work, probably not the case.

What are interesting problems? Deepseek and some other open source models do decently well in the gpu kernel benches. https://kernelbench.com/hard Most people think they’re working on “very hard” problems and they really aren’t

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#562

Post by Bnjoroge »

fraXis wrote: post #357
apatheticonion wrote: post #320

I use DeepSeek every day (via VSCode Insiders and Zed Editor). It's very affordable and, while it's slightly behind Claude (not sure how far behind Fable), it suits my working style well. I'm not using unsupervised multi-agent workflows and don't need a library of skills files - I'm writing most of the code and leaning on AI to help with mundane tasks - like; - generating types for APIs - generating boilerplate based…

How are you accessing their API? Through OpenRouter, or direct? Are you using DeepSeek v4 Pro? $2 seems a lot cheaper than my own experience accessing them through OpenRouter for over 100 million tokens, but I am using OpenRouter to access v4 pro.

Deepseek direct is atleast 2x cheaper than other providers serving it. Maybe their caching strategy is just significantly better or they’re subsidizing api pricing. I think it’s the former. $2 is closer to 50 million v4pro tokens in my experience

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#563

Post by wat10000 »

teravor wrote: post #384
wat10000 wrote: post #366

Earlier quoted context omitted.

That's fine, but what stops those Chinese nationals at Geely from messing with their cars now?

do you have evidence that if they wanted to do that they would not be giving orders to European or American engineers? as I said, financial fictions don't matter in matters such as these. if you have evidence that they have Chinese nationals working as engineers without oversight, you should submit it to the government.

“Give us the signing keys to the software update servers.” “No.”

You really think that’s how that conversation would go?

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#564

Post by Bnjoroge »

apatheticonion wrote: post #400
fraXis wrote: post #357

Earlier quoted context omitted.

How are you accessing their API? Through OpenRouter, or direct? Are you using DeepSeek v4 Pro? $2 seems a lot cheaper than my own experience accessing them through OpenRouter for over 100 million tokens, but I am using OpenRouter to access v4 pro.

I am using Flash and accessing the API directly via vscode insiders and occasionally Zed (it's buggy but I keep coming back to it because I want it to succeed). Unless you need enterprise multi-model management, I don't see the point in OpenRouter as it just adds cost overhead and you can just self-host an open router alternative (LiteLLM, Bifrost, etc). Running an LLM gateway locally is kind of nice as it allows you…

You can BYOK with openrouter and pay nothing up until 1mil requests/month iirc which is pretty generous for one person

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#565

Post by nonameiguess »

steelframe wrote: post #456
hodgehog11 wrote: post #389

Earlier quoted context omitted.

In terms of price-to-productivity, nothing will beat DeepSeek right now. For what you have described, all of the existing frontier models will perform well (probably about the same even). If you expanded the list to very hard research tasks, Fable was so far ahead of the others that it doesn't even deserve debate. If you are a researcher doing something involving scientific computation or mathematics that wasn't reje…

Makes me wonder what the value of a PhD is.

Depends on the field. There aren't really all that many research tasks that can be completed in a week. One of my favorite stories from the past few decades was the finding of accelerating expansion of space. It took three decades of scheduling telescope time, looking at exactly the right places in space at exactly the right time, to find enough supernovas to have data to even analyze. Not a whole lot of non-trivial science is just "read existing literature, think hard, and produce text."

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#566

Post by Bnjoroge »

epolanski wrote: post #545
apatheticonion wrote: post #320

I use DeepSeek every day (via VSCode Insiders and Zed Editor). It's very affordable and, while it's slightly behind Claude (not sure how far behind Fable), it suits my working style well. I'm not using unsupervised multi-agent workflows and don't need a library of skills files - I'm writing most of the code and leaning on AI to help with mundane tasks - like; - generating types for APIs - generating boilerplate based…

I can second this. Deepseek is great for mundane work day tasks. I use it via claude code, just pointing the api to deepseek. It's also not a clear "Opus 4.8 >> DS 4 Pro", I've done 16 tasks in 4 days across the two, and while Opus was indeed on average better, both models performed well being able to handle most of my workload. In fact DeepSeek was _significantly_ better on 3 task out of 16 and Opus was _significant…

Have you tried glm 5.2? It’s better than ds4pro and very close to opus4.6. Alot more expensive than ds4pro/mimo but significantly chesper than opus 4.6 for 90% of the quality

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#567

Post by chadgpt3 »

sergiotapia wrote: post #141

These bastards already prevent me from buying a BYD car, and a xiaomi phone, and they are adamant about me not using a chinese AI model. I hope they do not succeed.

I am using a Xiaomi phone. The hardware is fine, the camera is very good compared to the average, the software absolutely sucks, bootloader unlocking is not possible and it comes full of spyware and adware that you can't disable.

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#569

Post by MisterTea »

Tangurena2 wrote: post #551
ddxv wrote: post #313

The US is slowly becoming more like China. From talks of nationalizing companies to make US state owned entities to banning foreign competition. It's just so strange how you become the thing you fear.

I think America is becoming more like Russia, with a dictator who eliminates any billionaire who disagrees with His Dictatorship . The American groups who want to go back to the "glory years" of the 1950s are equivalent to the Russian groups who want to go back to the "glory years" of the Soviet Union.

This is exactly where we are headed: a has-been that sold itself down a river to a bunch of kleptocrats who only are about themselves and their sycophants.

Re: US holds off blacklisting DeepSeek, more than 100 firms deemed security risks

#570

Post by ge96 »

inigyou wrote: post #441
ge96 wrote: post #260

Earlier quoted context omitted.

If they had encrypted LLM compute then in theory wouldn't matter who/what is running it

Well they don't, and nobody has any idea how to build that, so why even mention it

It is being worked on this is just a random paper I pulled but yeah

https://arxiv.org/html/2502.01013v1

few of em actually

https://openreview.net/forum?id=PGNff6H1TV

Post reply on HN