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

DeepSeek V4 – almost on the frontier

simonwillison.net

Post reply on HN
Print view View original

391–400 of 420 posts

Re: DeepSeek V4 – almost on the frontier

#391

Post by sowild_fun »

taffydavid wrote: post #21

I tried deepseek v4 through open code at the weekend. I'm a daily Claude/Claude code user. I tried to build something simple and while it got the job done the thinking displayed did not fill me with confidence. It was pages and pages of "actually no", "hang on", "wait that makes no sense". It was like the model was having a breakdown. Bear in mind open code was also new to me so I could be just seeing thinking where…

Using a bunch of CLIs to work with DeepSeek V4, I've found that Langcli is the best fit for DeepSeek V4. For programming tasks, the cache hit rate is above 95%. Not only can it seamlessly and dynamically switch between DeepSeek V4 Flash, V4 Pro, and other mainstream models within the same context, but it is also 100% compatible with Claude Code.

I previously encountered the "reasoning content missing" issue when using opencode + deepseek v4. I don't know if it has been fixed now.

Re: DeepSeek V4 – almost on the frontier

#392

Post by mastermage »

onlyrealcuzzo wrote: post #262
wg0 wrote: post #9

Deepseek v4 Pro feels like Claude Opus 4.6 in it's personality but here's what I did find out about costs: I did cut loose Deepseek v4 on a decent sized Typescript codebase and asked it to only focus on a single endpoint and go in depth on it layer by layer (API, DTOs, service, database models) and form a complete picture of types involved and introduced and ensure no adhoc types are being introduced. It developed a…

> It obviously went through lots of files in both prompts but total cost? Just $0.09 for the Pro version. When people say that LLMs aren't worth it, it kills me. A lot of us, on average, make $100+ an hour. $0.09 is You can't even read the vast majority of prompt responses that fast. LLMs will continue to get better (I'm doubtful at previous rates, all indications are showing that progress is slowing and costs are in…

Very American centered POV

Re: DeepSeek V4 – almost on the frontier

#393

Post by theshrike79 »

gspetr wrote: post #200
deaux wrote: post #72

Earlier quoted context omitted.

I think Google, and likely Anthropic, indeed do honor the settings chosen by the user. For Google in particular it'd be very surprising if they didn't. That's also why both do everything they can to trick users into allowing it. OpenAI, I wouldn't be surprised if you were right.

You mean the same Anthropic, that wouldn't blink an eye at intentionally overcharging users hundreds of dollars just for having a HERMES.md file in a repo, would be above taking your data for... ethical reasons?

They also INTENTIONALLY gave people full refunds for that case.

Re: DeepSeek V4 – almost on the frontier

#394

Post by sagarm »

0xbadcafebee wrote: post #198
eckelhesten wrote: post #67

Earlier quoted context omitted.

At least that’s what they’re telling you. It’s a ”trust me bro” scenario. I’d rather use the phone home version (deepseeks own endpoint). The benefit is that I’m fairly certain that they actually host the model I’m paying for.

If you're not Chinese, and you start a company outside of China, and your whole pitch is "We run open weights and we have nothing to do with China", 1) why would send data to China?? 2) why would you risk your business to do a thing that makes no sense?

A fly by night operation created primarily for the purpose of collecting training data and corporate espionage will make whatever claims they think will get them the right traffic.

Re: DeepSeek V4 – almost on the frontier

#395

Post by naaqq »

gbgarbeb wrote: post #373
naaqq wrote: post #126

Earlier quoted context omitted.

Sorry, I was wrong here. I meant a single long session. And there’s no compression, the 1M context is only half used.

Then where did 200M come from? 200,000 tokens?

Not all read tokens are included in the context, many of the tokens are from read cache hits. I hit it many times so it grew to 200M. The number came from the API platform.

Re: DeepSeek V4 – almost on the frontier

#397

Post by flumpcakes »

Accacin wrote: post #293

I tried DeepSeek via chat, and gave it a rather simple question: "Can you tell me who was on series 8 of Taskmaster, and what's the general opinion about the series? No spoilers!" It told me amongst other things that Paul Sinha was diagnosed with Parkinsons, as well as who the winner was. Then I said, "But I said no spoilers!" And it apologised for telling me Paul Sinha was diagnosed with Parkinsons.

This is hilarious!

Re: DeepSeek V4 – almost on the frontier

#398

Post by spaceman_2020 »

cheshire_cat wrote: post #69

While the cost are lower than frontier models there are two factors that make DS4 Pro and K2.6 not as cheap as they might look. For DS4 Pro there's a discount going on for the official API, which sometimes gets overlooked and mixed up in discussions. Simon uses the full price in the comparison, so that's not an issue here. The other issue is that DS4 Pro and K2.6 often use way more reasoning tokens than the frontier…

Pretty much every AI model is also on discount - its just not explicitly stated

Re: DeepSeek V4 – almost on the frontier

#399

Post by knollimar »

spaceman_2020 wrote: post #398
cheshire_cat wrote: post #69

While the cost are lower than frontier models there are two factors that make DS4 Pro and K2.6 not as cheap as they might look. For DS4 Pro there's a discount going on for the official API, which sometimes gets overlooked and mixed up in discussions. Simon uses the full price in the comparison, so that's not an issue here. The other issue is that DS4 Pro and K2.6 often use way more reasoning tokens than the frontier…

Pretty much every AI model is also on discount - its just not explicitly stated

How does that hold true for open weight ones?

Re: DeepSeek V4 – almost on the frontier

#400

Post by jml78 »

tw1984 wrote: post #360
deaux wrote: post #73

Earlier quoted context omitted.

You're making completely unfounded assumptions about me. I use Chinese models myself.

Anthropic and OpenAI took your data, trained their model, and tell you "we are not going to tell you anything how we trained our models, we are not giving your the weights our models, you will have to pay us to access the model trained from your data". they took your rights and your data. Chinese labs took your data, trained their model, and tell you "this paper details how our models are trained using your data, her…

Hold up. Look, this is all shades of grey but saying Chinese labs all release open weights stuff is kinda crazy thing to say.

Right now they are doing that because they are still trying to catch up to Anthropic, Google, and OpenAI.

The moment they have the special sauce, they will shut it down and you won't be able to run their stuff anymore outside of them. Why do I say that? We already have the evidence in the diffusion model arena. All the chinese labs were pumping out open weights models for image and video, the moment they got to SOTA, they stopped doing it. Less and less is being released.

Chinese companies aren't doing open weights models out of the goodness of their hearts, they are doing it because it help their entire industry catch up. Don't get it twisted, this is very much a US vs China battle here. China wants to win and I am not sure how they won't. Deepseek is the first major large model trained on Huawei chips. It won't be the last and I am betting that China will make up for lesser performance of those chips with more manufacturing and power generation.

I am very bullish on China winning the AI war here. But I also am not naive enough to think that the Chinese companies is doing open weights out of wanting to make the world a better place or the goodness of their hearts. It undercuts the american AI companies.

Post reply on HN