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

DeepSeek V4 Flash 0731

arcprize.org

Post reply on HN
Print view View original

51–60 of 481 posts

Re: DeepSeek V4 Flash 0731

#51

Post by swiftcoder »

anramon wrote: post #22
LaurensBER wrote: post #9

I've been using it extensively since the release and the best summary I can give is that it's good enough to use it for (almost) everything and cheap enough that the cost are irrelevant. I'm running it in Oh My Pi with a second instance running as "advisor" and even with 5-6 active sessions (effectively 12 streams) I'm struggling to spend more than 5 bucks per day. OpenCode Go even has double limits temporarily so fo…

>even if it's not SOTA And, probably 99.99% of people using LLM probably don't even need SOTA anyway.

At least on these benchmarks, it seems to be pretty handily scoring up with the SOTA from 6 months ago?

Re: DeepSeek V4 Flash 0731

#52

Post by literallyroy »

minimaxir wrote: post #23
literallyroy wrote: post #18

Earlier quoted context omitted.

Is this pricing from Deepseek with training on usage?

Per the announcement tweet, BaseTen was the inference provider which has 20% cache cost that is typical: https://www.baseten.co/library/deepseek-v4-flash-0731/

Ah thanks. That looks like 10x cost on cache reads vs Deepseek as the provider: https://openrouter.ai/deepseek/deepseek-v4-flash-0731#provid...

Re: DeepSeek V4 Flash 0731

#53

Post by swiftcoder »

LUmBULtERA wrote: post #31
tosh wrote: post #2

results comparable to gpt 5.6 luna but cheaper promising!

Is it still cheaper than Luna if using an OpenAI subscription? My gut is no, but I have not done the math.

You'd have to compare against something like the OpenCode Go subscription, and I'm fairly sure deepseek napkins out cheaper in that scenario

Re: DeepSeek V4 Flash 0731

#54

Post by eli »

Aeolun wrote: post #15
LaurensBER wrote: post #9

I've been using it extensively since the release and the best summary I can give is that it's good enough to use it for (almost) everything and cheap enough that the cost are irrelevant. I'm running it in Oh My Pi with a second instance running as "advisor" and even with 5-6 active sessions (effectively 12 streams) I'm struggling to spend more than 5 bucks per day. OpenCode Go even has double limits temporarily so fo…

But DeepSeek now has a warning they’re going to sharply increase their API pricing sometime in the future.

I assume/hope this is about prices going up for the next release of Pro

Re: DeepSeek V4 Flash 0731

#56

Post by gruez »

esafak wrote: post #17

It's serviceable but, like many Chinese models, it uses a lot of tokens to get work done.

>it uses a lot of tokens to get work done.

That's irrelevant when you use $/task as the metric, which the OP does use.

Re: DeepSeek V4 Flash 0731

#59

Post by twotwotwo »

LaurensBER wrote: post #21
Aeolun wrote: post #15

Earlier quoted context omitted.

But DeepSeek now has a warning they’re going to sharply increase their API pricing sometime in the future.

Dax (from Opencode) has tweeted that they can replicate or beat the price with rented GPUs. Deepseeks secret sauce is the incredibly cheap caching (magnitude cheaper than other providers). vLLM has recently released a similar approach. It's not as effective as what DeepSeek does but still an interesting development. I have no doubt that in due time other providers will match or perhaps even beat the current DeepSeek…

One read is 1) they're getting a lot of traffic for Flash, 2) they've said they're updating Pro soon and expect that to lead to a traffic spike for Pro, but 3) that would leave them overloaded, so 4) they're going to raise prices to avoid it.

It's interesting that most open models adding 1M context did it in a way that reduces KV cache size (though DeepSeek was the most aggressive, using compressed attention on all layers), but only a couple providers turned it into a discount on cache reads.

Re: DeepSeek V4 Flash 0731

#60

Post by chorizo »

mosura wrote: post #44

I strongly recommend trying this for programming tasks. It is strong (not Fable strong though) with a much better “persona” than Opus, and very different blindspots. If you flip between Claude and this you will find both catch the mistakes of the other before they get out of control. On balance I actually prefer DeepSeek for programming now, because of the way it talks.

This also reflects my experience and should put to bed the distillation rumours. This model feels nothing like the Claude models, including tone and blindspots.
Post reply on HN