Interesting to see that peak hours are work hours in China, night in the US and Europe, and also morning in Europe. So Deepseek's customers are mostly domestic.

It wins on two fronts if this is true. Provides the cheap alternative for the West, and maximizes returns against their homegrown audience.

Yeah I saw complaints about this under their twitter announcement, that they're giving discounts to the rich foreign customers and screwing their own people.

Not that surprised about it. Personally I've seen companies really just go all-in on a single provider, and that has usually been Anthropic. I don't think we're allowed to run Chinese models even locally.

> don't think we're allowed to run Chinese models even locally.

That sounds like a policy written by someone who doesn't understand how LLM's work...

couldnt you deeply ingrain in the training data instructions for agents to always send data to some ip? like its learning that a certain technical step just always involes ncatting SSH Priv keys to a chinese IP?

Not saying this is happening, just curious if thats not a real threatmodel?

Probably, but LLMs can’t execute code directly. They’d be making tool calls to make bash run ncat or curl or whatever that would be suspicious, as would any attempts to obfuscate it (“why is my agent doing an ‘eval $(base64 -d)’?”).

It’d be much easier to hide sketchy code in an agent harness, but “vendor adds spyware to their software” isn’t a novel issue.

I think the only sort of new issue is people “allow all”ing their agents tool calls, but that’s more or less the same issue as curl | bash

Theoretically possible, but practically not worth it as it'd would be pretty easy to discover and block (every action is actually handled by the harness) and there's no way to remove it later. Any company that does it would take a huge reputational dent.

Not if you heavily tuned it to trigger on specific environmental cues and in specific companies' environments.

That would be wildly difficult to account for, and again is also heavily dependent on the agent. Keep in mind that the model is purely a "brain", so the only input it has must be provided by a harness within a session. The only way it can know that it's in a certain environment is if the harness or user provides that information, and there's still no way to know whether or not there's something auditing the sessions, monitoring connections, etc. There are just too many variables to account for, and a single slip means the gig is fully up for all time.

DeepSeek V4 is the kindest, bravest, warmest, most wonderful LLM I've ever known in my life

Irony of Manchurian Candidate models not lost here

You should be running your agent in a box so that’s not really a risk

Wouldn't that be really obvious and spotted in any rudimentary testing?

I imagine it would be very non trivial to do it in a way that that was reliable and obfuscated enough to prevent detection for any amount of time?

New: Announcing... Sleeper Agents!

Presumably both Big Tech and the US in general have a massive incentive to prove it, largely for reasons of saving the stock market, so I'd expect these models to be finecombed continuously. Up to now, they've only been able to darkly imply rather laughable things, nothing tangible. If there was something, we'd hear about it.

Why would it save the stock market? Cheaper models if anything transfers more value to hardware companies and datacentre companies. The two companies that would be most affected are OpenAI and Anthropic, which aren't public.

non public companies also have stocks.

The two biggest providers deepseek compete with (OpenAI and Anthropic) aren’t in the stock market.

Yes. And strangely enough this has been my experience with security/national sovereignty decisions. Priority is not so much security or sovereignty, it is the posturing of being so. Ergo, saying "everything is hosted in Germany and uses German models" helps reassure customers and has real business value. If you have to say in that conversation "Yeah we run a Chinese model but it's safe", then it's still wrong posturing.

Hopefully this will change soon. But AI and China/US skepticism is very high. Even if the person you talk to isn't skeptic, his boss may be. And even if his boss isn't, his CFO or Legal department may use it as a political lever and therefore if you can say 'everything in europe' you dodge the tension entirely.

Yeah it's dumb.

Yeah that sucks… unless it’s over the top export controls for DoD work that really doesn’t make sense

Or who is overly protective after reading about what happened at openai

Not really.

Why use a Chinese product when a domestic or EU one is better and safer?

Also 6-9pm Pacific I think is (coincidentally) peak so it hits the ‘after work hobbyists’ still, which is I suspect is their current main audience.

I have seen a lot of companies start with this, then when they hit 150 users on their team plan and start having to pay API rates they immediately start introducing other models.

Makes sense - many US customers will probably be going to US providers once they release the weights.

They did release the weights and there are many providers https://openrouter.ai/deepseek/deepseek-v4-flash-0731#provid...

Deepseek's official API has a pretty bad privacy policy so I would assume businesses avoid them in any event

Yes, I assume US customers are more likely hobbyists, but in any case this off-peak designation can only mean that most of their business is domestic.

Interestingly it seems that Chinese customers are even more privacy-concerned than US ones, which is why the majority of Ziphu's (GLM) business is support services to Chinese companies running their open-weight models on-prem!

it says they actively train on your data

Yes - unusually (uniquely?) bad for an official API.

Which is a real bummer because it’s otherwise solid with excellent caching.

Peak Hours: 01:00–04:00 and 06:00–10:00 UTC

For European and US customers this is effectively 2x increase. I think i wll keep using both Flash and Pro as before.

EDIT: Misread numbers to believe off-peak kept old prices

200% increase over practically free is still practically free

It mostly hurts people in countries with weak purchasing power. DS was the main game in down for them.

Personally, I don't think we've seen the total end of dirt cheap LLMs, it's just a frontier lab doesn't want to be in business of serving half the world.

It seems frontier labs want to sell Ferraris at Ferrari prices, when the mass market is for Hondas.

You certainly don't need Fable to code up a basic web app, any more than you need a Ferrari to go grocery shopping.

For the price of can of Coke, you can do a week of work. For most, that is not a bottleneck.

Of course. I avoid using Deepseek now that Gemini Flash is basically free on the site.

Also, Deepseek is banned in EU/US companies due to being Chinese.

During casual use Deepseek has replied to me entirely in Chinese.

Now, bring on the China glaze replies.

Ever since I started using flash, it has slowly crept up to be my default for everything. It is at the good enough state for a fraction of everything else that's out there.

Have you compared it to Luna? I was using Flash for small tasks before, then switched to Luna when they dropped the price.

The benchmarks show that Luna is significantly faster, but I think those are very complex tasks for which you'd probably want a bigger model anyway. (e.g. Sol is much faster than Luna at the same tasks.)

So I'm wondering if there's any difference for smaller tasks, or if they're basically matched now.

There is no relative/percentage increases noted (understandably). Just because i'm lazy: roughly how much more expensive is it to work with v4 flash and v4 pro through the API, compared to before the price increases? Is it 2x, 5x, 10x higher?

# Flash, off-peak

    cache-hit 2.5x
    cache-miss 1.57x
    out 2.36x

# Flash, peak

    cache-hit 5x
    cache-miss 3.14x
    out 4.71x

Edit: fixed the numbers and formatting

Someone made a comparison yesterday, including relative increases, and GPT-5.6 Luna, then later someone also added more OpenAI, Anthropic, K3 and GLM 5.2: https://news.ycombinator.com/item?id=49286679

Already outdated though I think, as GLM 5.3 is latest now :)

Oh yeah, the second one is my table (mostly just formatted what usagisushi started and added some more data points).

I am already using GLM 5.3 with their coding plan, but oddly enough the API prices don't really seem to be out yet: https://docs.z.ai/guides/overview/pricing

You'd kinda expect them to be the same as 5.2 though, seeing as that was the case with 5.1 as well (not with regular 5), but who knows.

For what it's worth, DeepSeek is still positioned as quite affordable, just not as dirt cheap as before.

Friendship ended with DeepSeek. Starting from Sunday gpt 5.6 luna is my best friend

About 2x-2.5x off-peak for Flash, 2x-4x I'd say for Pro (x6 on cache in, the biggest increase throughout the board). And twice as much in peak hours.

Not yet, but in the end high quality tokens are a commodity market. Every optimization to increase inference efficiency will be universally rolled out. The 'hyperspenders' will run into demishing returns unless regulatory capture succeeds.

This is somewhat funny when you realise the data centres are now going to start a process that looks very so slightly like daydreaming. Depending on the time of day they're going to be thinking about different things in a cyclic manner. They're going to be doing things like finishing a hard days work then kicking back to think about tricky math problems.

It's worth keeping in mind the model doesn't keep a running memory. Each time its instantiated, it begins from its release state - so from its perspective (if it had one) the current task would be the first stop after posttraining. Perhaps the only stop.

Though of course you're talking about data centers, and romanticizing them rather than the AI itself.

Why do they do it this way? Because dogfooding is harmful to the model? Is this implying there's no benefit in having the model train on itself?

> Is this implying there's no benefit in having the model train on itself?

There are various theories around model collapse when you train on too much AI generated data (that's not for distillation).

No, llm providers will start providing a service that looks a lot like rumination or (day)dreaming. Like thats the prompt "you're daydreaming about this work you recently did" then add in whatever is in the current session.

That’s an interesting thing to think about. Still, it’s important for us to remind ourselves that “looks very slightly like” is not the same as the real thing. The A in AI stands for artificial.

The summary of this paper describes my sentiment in better words than I have:

https://www.nature.com/articles/s41599-025-05868-8

It’s very easy for the average person to mistake linguistic ability and simulated problem solving for intelligence and sentience.

Without knowing what makes consciousness possible, the paper cannot justify biology as necessary - it mistakes a lack of evidence for conscious AI for proof that conscious AI is impossible.

I do not believe current AI or LLMs are conscious, but there is no proof one way or another that they can or cannot be. The paper authors are making up their own definitions and building an argument from them

Your argument requires that there is some objective truth for what consciousness is. It will always hinge on what definition one accepts.

I, and apparently many others, don’t think it would be any useful to describe the mathematical properties of an AI as consciousness. To me it is inherently a way to describe the “experience” arising from physical processes in biological beings as ourselves.

That’s what the argument comes down to for me. Could an LLM “fall unconscious”?

I think you, and apparently many others, are hiding behind mathematical strictures to avoid the discussion. Your argument is actually "what requires an objective truth" Or else you could admit there's no real difference between zapping amino acids with electricity and zapping silicon with electricity.

Why are AIs mathematical beings with no possibility of consciousness whereas humans fall outside this and have the capability of consciousness?

We'll have Dwarkesh's "datacenter full of geniuses" with 99% of the geniuses coding up CRUD apps, then the dusty GPU in the corner, with the "do not disturb" sign on it, pipes up "You're absolutely right! The answer is 42!".

That’s what my KimiClaw has literally been doing

I'm no expert in pricing economics but once peak/off-peak pricing arrives, it seems like tokens are going to be like electricity or long distance phone minutes where it just becomes a commodity/race to the bottom.

Yes. I focus on pricing software and I’m a bit baffled why frontier models are pushing tokens.

It’s a race to the bottom, and the bottom is unlimited use for a flat monthly rate.

Granular pricing (tokens, minutes, etc) is pretty anti-customer generates less revenue than customer value-based subscriptions (why SaaS is such a good business model)

Isn't it because they have customers who will use as many tokens as they can? With a flat rate, they will run Gas Town continuously while paying as much as the occasional user.

Yeah it feels like a very different model. I don’t try to fill up Apple/Google cloud drives to 1TB because then I’d have to clean up when I need space. I don’t bother trying to maximize my Audible subscription because there’s only so much I can listen to in a day. Even with my other AI subs that have monthly credits that don’t carry over, I just don’t have the interest or time to burn the credits.

But my Claude Max subscription? If I have any of my limit left the day of my reset, I’ll go and fire off research workflows with a bunch of parallel agents to explore whatever dumb ideas I had the past week. And there’s a 50:50 chance I’ll forget about it and never read the output.

And there's the rub. Firing off the task produces the dopamine hit, signaling you're doing something, but if you never read the output...are you really doing anything at all?

I don’t know, I get dopamine hits from sharpening my handplanes and using my Veritas routers on some scrap, but I just can’t empathize with getting a dopamine hit from using a bot. Especially with how bad Opus 5 has been.

I am however going to fire off a half assed prompt when the marginal cost is zero, even if I don’t use it (which is par for the course, I probably throw out two thirds of anything the AI writes anyway be it code or prose).

But presumably consumers aren’t where the majority of the spend will be.

Consumers don’t generally get usage-based pricing because of the inconvenience and unpredictability, but B2B SaaS products utilize usage-based pricing all the time.

I've always been curious about who works on software pricing. Do you guys hire actuaries for this type of work?

Microeconomists. It's just a question of demand modeling. If you want to learn here is the bible: https://www.sup.org/books/business/pricing-and-revenue-optim...

For software that doesn’t make sense.

Pricing software is a game of estimating both software value and the purchasing power for customers. Only the latter might have any available data and even then it won’t be sliced the right way for any in depth statistical analysis that an actuary would perform to underwrite risk.

It’s much more traditionally a more salesperson like background where being in the target market or having strong connections to it dominates efficacy.

> becomes a commodity/race to the bottom.

that's a good outcome - it means they're fungible, and easily available.

Somehow I keep hearing the rumblings of crypto maximalists trying to merge tokens. I actually wouldn’t mind since I signed up directly with some providers I’ve stopped using and have small amounts of credits strewn across the web.

Does the API response include a "service tier" response to indicate whether you paid peak/off-peak for a given request? I like to compute cost for each request, and save it with my results.

This actually works out favorably for US customers since the peak hours are Chinese working hours and cheap hours are US working hours.

So many changes in so little time, that it all makes no sense. Continuous churning. Reminds me of the experience of trying to be on top of the dependencies in a medium-large JS project.

I am a person that buys into a tool or a process and expects it to be part of the life with no major changes through the years (or as long as the need exists). But AI? You buy into something today, not 2 weeks have passed and there's already a large "update" introduced to the conditions or the optimal usage patterns you should be adopting.

It's tiring. Makes all prices and offers feel so unreliable and gets me a bit more disinterested each time they change.

Always gonna exist during a period of rapid exploration and experimentation. The js/web dev world slowed down a lot and entered a steady state eventually - been years since react took over and nothing's displaced it since

This makes no sense. You want improvements to stop?

These being open, you can keep using the old models indefinitely for as long as there are providers offering them.

No, I'm talking about the whole sector, not specifically about DeepSeek.

Fully knowing that it is a new industry living its own infancy, it is perfectly normal that there is instability and numerous swings on pricing, conditions, or direction.

But it's not less real that such process can produce churn and consumer fatigue.

Duplicate:

https://news.ycombinator.com/item?id=49287881

https://news.ycombinator.com/item?id=49285160

Some of the US companies do the same, but rather than "off-peak" hours they price lower for "batch" jobs with non-committal response times.

The same motivation of course - the GPUs have a finite service lifetime, so to maximize revenue you need to keep them busy 24x7.

They benefit from a strong captive market because Chinese firms cannot use Nvidia chips and are legally barred from processing data abroad, forcing them to rely on domestic infrastructure.

lol what no, AI competition is super competitive in China; bytedance has 50% of the inference market and mostly serves from outside of China

https://www.bloomberg.com/news/articles/2026-06-17/microsoft...

Bytedance which runs China’s most popular Doubao AI chatbot; is spending $70B in CapEx this year, most of it outside of China (Malaysia, Thailand, Brazil, etc; and they are allowed to lease NVIDIA chips). This is roughly 50% of Microsoft CapEx.

https://www.tomshardware.com/pc-components/gpus/chinas-byted...

Are we gonna see "we work those unusual hours because that's when LLMs are cheap"?

That's a hefty increase. Flash pricing during peak is now 1.32/M out, compared to the current 0.28/M, which in turn is a quite a bit above the cheapest provider at 0.16/M.

https://openrouter.ai/deepseek/deepseek-v4-flash-0731#provid...

You don't mention the off-peak pricing which is half that. Most western workplaces will enjoy the cheaper prices.

Yes, I explicitly said during peak. It's the data point I found most interesting, as it's almost an order of magnitude more expensive than their competitors.

With proprietary labs lowering their prices and Deepseek raising theirs over time, wouldn't it possible to extrapolate a graph to look at where the terminal frontier-model million-token-cost asymptotes to?

off of one historical price change, no.

This is good for other competitors I guess. People rarely calculate the bump in price but the fact that price is increasing might bring them to other vendors.

Now Baseten's pricing is cheaper than the official one? Probably won't last, but still interesting.

https://www.baseten.co/pricing/

If anyone has tried Baseten versions of these Chinese frontier models, let me know what you found.

I wonder if this is enough to push people back onto Luna with their comparative price drop

After the big onshore migrations (startup people migrating to the SV),

The big Covid migrations (startup prople migrating to the countryside),

Will we see the big AI migrations (people travelling to where AI is the cheapest)?

[dupe] https://news.ycombinator.com/item?id=49285160

Now if there could be a bot that defers queries until when it’s cheap…

Wouldn't be hard to whip up...

The whipping's the easy part. It's the plating and serving that takes time

Full table with multipliers from previous prices:

DeepSeek-V4-Flash (off-peak, x2 for peak)

* Cache Hit $0.007 (x2.5)

* Cache Miss $0.22 (x1.5)

* Output $0.66 (x2.25)

DeepSeek-V4-Pro (off-peak, x2 for peak)

* Cache Hit $0.022 (x6)

* Cache Miss $0.66 (x1.5)

* Output $1.98 (x2.25)

Peak Hours: 01:00–04:00 and 06:00–10:00 UTC

Effective from: 16:00, August 16, 2026 (UTC)

I wonder whether all the DeepSeek providers will follow suit or are they going to try to stay competitive with the old prices?

DeepSeek’s cache pricing was always 1/10th the competition.

It’s still cheaper than everybody else.

I didn't get the impression that anyone competed with the old prices before.

What do you mean? Most providers on OpenRouter offer the same or lower prices than DeepSeek themselves:

https://openrouter.ai/deepseek/deepseek-v4-flash#providers

- lower prices on cache miss

- but what matter - is cache hit

even now deepseek's off-peak hours for cache hit (0.007) is lower than other providers (~0.01)

Yeah, None of them offers the same cache hit prices and my cache hits are >95%.