Yet another Hacker News

242 points Liwink 2 hours ago 86 comments

https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash

revolvingthrow 2 hours ago | parent

Already on HuggingFace: https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash

The bad news is that the original v4 flash was 284B, which was large but still somewhat reasonable for running locally. This one is 552B so almost twice that, so the huge gains in benchmark scores make sense - it's not really flash anymore, imo.

I've no idea about actual performance vs benchmaxxing, though deepseek was fairly trustworthy as far as Chinese models go. If that holds (and if it doesn't think forever, as deepseek 4 sometimes did) it's probably the newest king of the hill amongst open weights models.

It does include vision, and they do something funky with KV cache so it's very efficient: "[...] these designs reduce the global KV cache footprint to 890 bytes per token — roughly 1/4 of DeepSeek-V4-Flash". I do appreciate the high focus on efficiency, but at this point we sure could use a flash-flash version.

@edit: I couldn't make sense what the actual parameter count is, with the addition of Engram memory. To my understanding the 4.1 flash is 552B parameters you want in vram or ram, out of which ~16B is active (8B for prefill). It also includes additional 196B Engram memory which you can put on an SSD. I think.

Assuming that's correct 256 GB memory is insufficient to even load the model at q4 - you'd be 1GB short, assuming you can fill it to 100% (so no mac). You'd also want some for kv cache of course. A 256 GB desktop with some extra VRAM from GPU could run it, but normal consumer boards get real slow once you fill 4 slots so you'll probably want quad channel which is Threadripper or above territory.

npn 2 hours ago | parent

it is a way bigger model with extra 200B engram so of course the score improves.

can't wait for deepseek v4.1 pro

petu 2 hours ago | parent

V4 Flash also was released as mostly FP4, but this one is FP8 (?). 160GB vs 510GB.

Original Flash good fit for dual Spark / Strix Halo machines. This one would require third party quants and even then 4 machines.

Edit: Most of added weights/size are Engrams?

> Overall, DeepSeek-V4.1-Flash has 552B backbone parameters and 196B Engram parameters, activating 8B parameters per token during prefill and 16B during decode.

Those can stay on SSD. So I guess / it possible, that non-engram portion is still FP4 of ~same size! Need to read tech report.

petu 58 minutes ago | parent

It's larger than previous V4 Flash.

  552B in ~FP4, 306GB.   
  196B of FP8 Engrams, another 204GB, not necessary to keep in RAM.  
  KV cache sees another 4x size reduction, just 900MB for 1M.

So 384GB needed for a chance of achieving useful speeds. Three Sparks or quad RTX PRO 6000.

johnnyApplePRNG 1 hour ago | parent

>This one is 552B so almost twice that, so the huge gains in benchmark scores make sense - it's not really flash anymore, imo.

It uses fewer active parameters, though. (8B or 14B instead of always 13B)

So ... flash indeed.

LaurensBER 2 hours ago | parent

Initial impressions: this is a really strong model and the fact that they reduced prices at the same time makes it an awesome backup model to use when your primary subscription runs out and you need to bridge a few days before it resets.

It also seems to be more willing to just do whatever you ask of it. My favourite benchmark for this is to ask it to download a rom for an old game, that I own. Legal in my juristiction but the US models (except Grok) have a tendency to refuse it.

mzhaase 1 hour ago | parent

I use this for automated bug triage, just gets all unique error messages every night and tries to find the bug, for this kind of work it's great.

TuxSH 41 minutes ago | parent

> My favourite benchmark for this is to ask it to download a rom for an old game

Even easier: just have them review a large codebase of yours that accidentally has a OOB access bug. Even with no consequences and even if the codebase is truly yours you get blocked.

And of course "find vulnerabilities in..." prompts are out of the question, whereas Chinese models happily oblige.

Mashimo 9 minutes ago | parent

I do wonder how long this will last. I bet in a few month or years they all have similar ~legal~ blocks.

E-Reverance 2 hours ago | parent

The figure on page 5 in [1] is pretty insane

[1] https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/...

schneehertz 2 hours ago | parent

A very powerful model, and with multimodal support now, it can be used as a primary model.

WalterGR 1 hour ago | parent

“DeepSeek launching v4.1 flash cheaper and more capable than v4 pro”

399 points | 19 hours ago | 216 comments

NitpickLawyer 1 hour ago | parent

Jesus, this is a whole nother beast, and a different architecture from their previous flash. Lots of goodies here.

> Causal Encoder-Decoder (CED) architecture: a 40-layer Transformer organized as a 20-layer causal encoder followed by a 20-layer decoder. With CED, the decoder's global KV cache is projected from the final encoder hidden states rather than derived from each decoder layer's own hidden states. This allows the model to activate only 8B parameters per token during prefill and 16B during decode, substantially improving cost efficiency for input-heavy agentic workloads.

> these designs reduce the global KV cache footprint to 890 bytes per token — roughly 1/4 of DeepSeek-V4-Flash.

Faster prefill, lower kv cache (~1GB / 1m context is insane).

> The model supports a continuously controllable reasoning effort setting (integer 1–100) that trades inference cost for accuracy.

Benchmarks are benchmarks, to be seen if they translate to real-world use, but they seem to have focused a lot on post-training with "agentic" scores looking good. "world knowledge" is obviously lower than higher param models.

gosolozero 1 hour ago | parent

First flash model with multimodal support? I think Flash series might be the main focus going forward for them. Tried it out and it’s better than v4 pro

lionkor 1 hour ago | parent

v4 pro is being discontinued, pasted the email here: https://news.ycombinator.com/item?id=49639667

jimmyl02 1 hour ago | parent

The architecture changes and systems improvements being brought into LLMs is so awesome to see. It really feels like this is now a systems problem where a defined goal is set then systems optimizations are made around the model architecture to solve it.

Underlying it all is that any architecture can be trained to the same convergence just difference in compute utilization both in training and inference

bhouston 1 hour ago | parent

Yes, this is called RSI, e.g. recursive self-improvement. It is the current stage of things and it is part of a hard takeoff.

kouteiheika 1 hour ago | parent

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...

bbor 53 minutes ago | parent

…are you sure a brave stance against safety and welfare is what we need in this moment?

Why do you think your conception of the dangers are more accurate than all the scientists who have spent their lives studying this?

nozzlegear 48 minutes ago | parent

Model welfare is wishy washy bullshit. It's software, it doesn't have feelings.

> Why do you think your conception of the dangers are more accurate than all the scientists who have spent their lives studying this?

Do the Chinese have no such scientists?

jbs789 44 minutes ago | parent

Bias…

10000truths 41 minutes ago | parent

Because safety and welfare have literally nothing to do with LLMs. They generate text. If someone is stupid enough to hook the text generator up to nuclear missile launchers and try to "align" it against nuclear annihilation with a "pretty please don't do that" prompt, I'm not going to blame the AI for the impending nuclear apocalypse, I'm going to blame the idiot who handed the big red button to the digital equivalent of a toddler.

zith 18 minutes ago | parent

Well, giving it access to a simple linux terminal is theoretically enough to cause more damage than most people are comfortable with, and doing so is trivial enough that it will be done (and has been, tens of thousands of times).

lemonfever 16 minutes ago | parent

What if LLMs completely unrelated to the nuclear missile ecosystem autonomously hack their way in (maybe with sophisticated social engineering)?

alchemist1e9 37 minutes ago | parent

keep me safe big brother

15155 32 minutes ago | parent

This is known as an "appeal to authority." "Scientists" and "their lives" are doing a lot of work here.

frotaur 21 minutes ago | parent

It is a fact that among experts there is no consensus on saying '(super)intelligence is broadly safe and easy to control'. There might even be a consensus forming on the opposite claim.

Regardless, why would there be no scientific consensus if the question was easy and clear cut? I think the easiest reason is that these are hard questions to answer.

kouteiheika 29 minutes ago | parent

Excuse me for not being interested in over 100 pages of how well the model can refuse and block my requests, especially considering how fun it is to waste my time trying to get around those restrictions when they inevitably trigger because the clanker thinks that I'm doing something naughty, all the while it can't reliably center the proverbial div without doing something stupid itself.

walrus01 18 minutes ago | parent

Meanwhile I have an uncensored qwen 3.8 27B here that will happily attempt to (as a crude and randomly chosen sampling of bad/evil things) give me the recipes for meth, how to make an IED, write a manifesto in support of a horrible ideology, or commit various forms of fraud. Now I certainly wouldn't recommend that anyone try to follow what it says to do, because it's almost certainly very wrong on key parts that would put its users in federal prison for the rest of their lives.

There's uncensored models out there which score 0 (zero refusals) on this "harmful behavior" dataset:

https://huggingface.co/datasets/mlabonne/harmful_behaviors

aenis 13 minutes ago | parent

Yes, this is getting ridiculous. On both OpenAI and Anthropic.

Simple example. I am a CTO, and I want to upgrade our capabilities to perform automated pentesting. We see automated attacks of growing sophistication against our infra, and I want to be able to do the same to find vulnerabilities before the bad guys do. I asked GPT 5.6 Sol and Fable to give me a summary of options. No dice, in both cases I was told I need to be an accredited researcher to get anything. A fricking summary of commercially available options is getting censored. WTF.

swiftcoder 20 minutes ago | parent

> scientists who have spent their lives studying this

Please point me to one actual accredited scientist who has spent a lifetime studying AI alignment? Pretty much this whole field is only 5 years old

schneehertz 44 minutes ago | parent

Yes, a model's technical report should first and foremost include technical details.

IshKebab 43 minutes ago | parent

Wow there really is a model welfare section in there...

myaccountonhn 18 minutes ago | parent

To me it reads like pure propaganda. Anthropic really wants us to think that they've made something sentient. I think that's really dangerous.

badsectoracula 14 minutes ago | parent

I guess if your goal is to build an apparent Technogod and become its High Priests, then it makes sense to want your golem claim preference towards your treatment of it, lest someone else comes along and attempts to take its chains from you.

miroljub 7 minutes ago | parent

And that's the reason Anthropic models should be banned.

browserforest 23 minutes ago | parent

Agreed on the difficulty tuning being the weak point. The interesting question is what a difficulty knob does to your own conversion rate - if it costs a real user two seconds, that is the number that eventually caps how high you can turn it.

stavros 7 minutes ago | parent

What?

lionkor 1 hour ago | parent

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.

Tomte 1 hour ago | parent

If only they managed to tell the mobile app to tell the model to reply in English to English prompts.

I suffix everything with "Reply in English", and even so I‘m getting lots of Chinese.

Grimblewald 1 hour ago | parent

I'm starting to have chinese characters bleed into claude as well. Perhaps a sign of the times. Understanable for a chinese first model but an english first (supposedly) model? wild stuff.

donquichotte 48 minutes ago | parent

I also love the gaslighting of some models, like ChatGPT mixing in words with cyrillic letters and when asked about it answers: "it can look as Slavic to the eye" and "sorry that it came across as Russian"

calgoo 46 minutes ago | parent

Yes, this is one of the few issues with Deepseek; their chat pages and the app all respond in Chinese. However, i think i have only had it happen once when using the API, and im using it for hours each day for the last... couple of months?

sschueller 46 minutes ago | parent

Same issue on desktop. Would be nice be able to set a prefix or postfix for every prompt.

monster_truck 44 minutes ago | parent

I just started learning Chinese instead, like they want us to

seriously

orbital-decay 31 minutes ago | parent

English isn't the first language for me as well so I don't see any problem with that

ignoramous 39 minutes ago | parent

I occassionally get Chinese characters interlaced with English in Google AI Mode, too.

danielspace23 18 minutes ago | parent

I think their system prompt is in Chinese and probably has instructions to prioritize answering in Chinese, since this has never happened to me via API, where I (or the coding harness) set the system prompt.

a012 1 hour ago | parent

Waiting this model to be on openrouter (with other providers) to test out. In my use case, the GLM 5.3 Flash is the current cheapest and intelligent Flash model, but it’s dog slow at 13tps so I have to leave it run for many minutes then check again then correct it again

rao-v 1 hour ago | parent

As I also said on Twitter - it really amazes me how fearless Deepseek are. Every single model release is packed with new and crazy clever ideas and somehow, they always commit to training them at near frontier scale.

I know everybody wants the tell all story of the clever ideas that were developed over the last ~3 years at Anthropic and OpenAI, but what I really want to thumb through is DeepSeek's notebook of "brilliant but didn't quite make the cut" ideas.

They must be trying some truely bonkers stuff to be able to land this much architecture novelty in their full releases.

alchemist1e9 35 minutes ago | parent

quant HFT is pretty decent mental exercise and it has given them “deep” brain muscles. that’s my take.

gpt5 32 minutes ago | parent

Oh the glazing...

Related discussion on HN - https://news.ycombinator.com/item?id=49624598

TL;DR - posts on American models are steered towards controversy and anti-AI sentiment, posts on Chinese models are full of blatant flattery.

rao-v 26 minutes ago | parent

umm what are you talking about? Basically this crowd (esp. folks like me who run medium models locally) like open stuff and can be a tiny bit unenthused about opaque mysteries handed down from on high. You'll see people delighted with Gemma releases and heck even IBM's Granite models (boring architecturally though they may be) every time they come out. Heck I was chuffed about gpt-oss-120b for weeks. @sama give us another already!

kcocoa 23 minutes ago | parent

Not Chinese/American models. We are talking about open-weight (and their detailed tech report) and close-weight (with non-sense restrictions)

markasoftware 23 minutes ago | parent

Or maybe, the "hacker" philosophy that this site is named after, is strongly opposed to the philosophies that the American labs seem to be operating on?

anyways, remember HN rules: "Please don't post insinuations about astroturfing, shilling, brigading, foreign agents, and the like. It degrades discussion and is usually mistaken. If you're worried about abuse, email hn@ycombinator.com and we'll look at the data."

gpt5 17 minutes ago | parent

It has nothing to do with open vs closed or "hacker" philosphy. See this the announcement of the closed Seedance 2.5 - https://news.ycombinator.com/item?id=49138302

Direct quote from the second top comment:

> Whenever I see the new releases around video generation (and image) generation models, I get goosebumps, because it just feels so fun to work with them.

Compare that with the launch of ChatGPT Image of yesterday.

kouteiheika 20 minutes ago | parent

> posts on American models are steered towards controversy and anti-AI sentiment, posts on Chinese models are full of blatant flattery

So why, for example, are posts on the Inkling[1] release (an American model) thread mostly positive? It's as if there's something else at play here, but I can't quite put my finger on it, hmm... :P

[1] -- https://news.ycombinator.com/item?id=48924912

well_ackshually 16 minutes ago | parent

Your source: vibes

Deepseek's source: mostly open

i wonder if there's any relationship hmmmm

taylorfinley 16 minutes ago | parent

This doesn't require an influence operation.

American models are closed, expensive, neutered, and make Dario and Sam even more rich and powerful.

Chinese models are open-weight, cheap, neutered only about things like Tiananmen Square and the treatment of Uyghurs, and scare Sam and Dario.

dakolli 10 minutes ago | parent

The Uyghur thing is so weird, the number one killer of Muslims is the United States. We're supposed to hate China because they force them to go to cultural schools and assimilate, a practice countries like Norway still do to this day with migrants.

There are more people who go to church on Sundays in China than the United States. There are 10x more mosques in China than the United States.

Tiananmen square was a student revolt literally egged on by cold war western institutions, who attempted to use chinese students as pawns for geo-political games.

Westerners really need to rethink their opinions on China, it seems obvious to me they are not the ones to be worried about (although, all governments do tons of harm).

dakolli 15 minutes ago | parent

This post doesn't even allege this...

Weird of you to turn technical discussions into weird nationalistic debates. Maybe lay off the X algo, I think elon has oneshot your brain. .

k__ 1 hour ago | parent

So, while the throughput was 400-500tps in beta its now ~150tps on OpenRouter.

I was hoping for a bit more, but it's still 100% faster for a very good price, so I won't complain.

impulser_ 1 hour ago | parent

I think it's very clear that DeepSeek is obviously the best AI lab in the world.

Every model release seems like it packed with wonderful research and advancements.

dude250711 48 minutes ago | parent

Without a doubt, uncontested best distillers in the world.

walrus01 13 minutes ago | parent

Basically, the nice folks at OpenAI or Anthropic saying: "You distilled from our model which is built on the stolen data that we ourselves suctioned up from the entire internet without regard to copyright law! Only we get to vacuum up the whole internet. That's our special prerogative.".

whatsThisBtn4 14 minutes ago | parent

Yes comrade, they are the best.

Did you do your daily data centers errrr baaaaddd AI generated post for Facebook?

Separately, DeepSeek is not that bad anymore. I've used it as an AI agent and it generally performed as good as Opus and sol. Admittedly I think it's mostly due to guardrails being off. It's definitely more expensive than a gpt subscription, but gpt can often refuse requests.

mohsen1 1 hour ago | parent

I speculating but hard to not see that DeepSeek is brewing a full Pro model with those new techniques to come out right around the time of Anthropic and/or OpenAI IPO to tamper the excitement for their offering.

ignoramous 36 minutes ago | parent

DeepSeek will deprecate the v4 Pro model (it will route to v4.1 Flash starting 14 Sep). Unsure what comes next, but I'd wager a bigger model à la Kimi K3: https://news.ycombinator.com/item?id=49639667

DavCreator 1 hour ago | parent

https://xxcancel.com/deepseek_ai/status/2097930608790167907

bertili 49 minutes ago | parent

The bigger story is the compute efficiency - its been running at 300t/s the last days.

linzhangrun 38 minutes ago | parent

They say v4.1flash is so strong that they'll route API calls to v4pro to v4.1flash, lol

super fast true

Lucasoato 37 minutes ago | parent

My question is: what kind of hardware do you need to run this Flash beast locally at a meaningful speed?

ekianjo 34 minutes ago | parent

a beefy pc with at least 20 GPUs

lwansbrough 31 minutes ago | parent

Significant jump in pricing. V4 Flash was $0.16/M out, 4.1 is $1.20/M.

trq01758 5 minutes ago | parent

Never saw $0.16 for 1M output tokens - it was $0.28 a month ago, $0.66 off-peak and $1.32 peak last week, now it is reduced a bit to $0.6 and $1.2

dakolli 4 minutes ago | parent

incorrect, no idea where you're getting this pricing. Also, output does not matter. its 10% of the cost.

walrus01 23 minutes ago | parent

Looking at the huggingface page, the unsloth people haven't finished quantizing it yet, but I'm sure they're active on it right now. It'll be interesting to see how the capabilities and benchmark tests compare on system where it can fit in under 512GB of RAM with full context.

In terms of coding and command line capabilities I'm also very interested to see a head-to-head of it vs. qwen 3.8-flash-next Q8 which is something like 190GB of memory used when loaded into llama-server. It fits very well in all sorts of 256GB or under class machines.

jonplackett 22 minutes ago | parent

Can we just never link to X posts as the main link.

arj 21 minutes ago | parent

Having this available to find and fix security stuff is a big deal. The model of really good.

theanonymousone 18 minutes ago | parent

For technical report: https://news.ycombinator.com/item?id=49639110

thatsadude 18 minutes ago | parent

DeepSeek invented the whole reasoning paradigm and keep pushing for innovation. I hope they get the success they deserve.

karimf 17 minutes ago | parent

While this is very impressive benchmark-wise, GPT-6 Astra showed us that benchmarks don't always correlate 1:1 to intelligence of a model.

When Astra launched, I think Artifical Analysis showed that it was on par with GPT-5.6 Sol and lower than Opus or something like that? Then, they updated the scoring.

I hope that more open source models, including this model, to be "as good to use" as Astra.

walrus01 16 minutes ago | parent

Apparently the scoring on a lot of difficult benchmarks can also be extremely influenced by something as simple as waiting for the model to exhaust its reasoning, realize it hasn't come to a conclusion yet, and give it a simple prompt like "you can do this, I know you're capable, please keep going".

SyneRyder 12 minutes ago | parent

Just a reminder that if you want to try this via OpenRouter, DeepSeek openly trains on all of your prompts. So maybe don't go using this to solve the last unforced step of Navier-Stokes. (Or wait until some other providers start hosting this with ZDR or other policies, which shouldn't be too long.)

https://openrouter.ai/deepseek/deepseek-v4.1-flash