Top
Best
New

Posted by nateb2022 20 hours ago

Kimi-K3 on HuggingFace(huggingface.co)
Related: Kimi-K3 Technical Report [pdf] - https://news.ycombinator.com/item?id=49070985
1309 points | 515 commentspage 6
martindelophy 16 hours ago|
The VRAM, power, networking, and operational requirements put it beyond the reach of many enterprises.
padolsey 17 hours ago||
There's no going back on this. This is putting a very capable intelligence in the hands of the masses. Private companies in the US are aching for Trump's protectionism but it'll do nothing. The hardware needed to run this is ofc prohibitive, but actually putting it out there feels like a 'RSA source code on t-shirt' moment for humanity.
edot 14 hours ago||
No, luckily private companies in the US are aching for this to NOT be banned. NVIDIA, Microsoft, etc. just released that letter. We’re saved from the trillionaire companies (OpenAI, Anthropic) by the other trillionaire companies acting in self-interest (hosting and hardware).
efficax 12 hours ago|||
I don't know if the masses can quite afford the 500k in GPUs you need to run this
supjeff 11 hours ago||
a "moment for humanity"? as if this shit isn't going to generate 99% slop at the cost of all we have left as a species?
asdewqqwer 4 hours ago|||
The only uniqueness of humanity is intelligence. If there is really AGI, human definitely is more close to it then other animals.
api 11 hours ago|||
Sometimes I’m not sure who is more unhinged: the total AI kool aid drinkers who think this will make us all into immortal demigods (or take over the world as it goes “foom”), or the AI doomers and haters who exaggerate everything potentially negative about it and react to it the way a 1980s Christian fundamentalist reacted to rock music.

It’s a new fundamental innovation in math and CS that allows large scale lossy compression of natural language another data formats in a way that is semantically queryable and cross-referenceable. It also manifests some form of emergent intelligence, likely evidence of the long posited link between intelligence and data compression.

The tech is awesome. It’s one of the coolest things I’ve seen in over a decade. The industry is kind of shitty, which is not unusual. The discourse around it is almost universally insane, crazy people arguing with crazy people.

Oh and get off the AI eco bullshit train. Look up the energy cost of AI queries vs driving or running a home air conditioning system. Feel bad about using AI? Skip that DoorDash order. You probably just saved the energy of 1-2 days of heavy Claude Code use.

mwcampbell 8 hours ago||
To steelman the haters, I think their view is that the industry is so uniquely shitty that it's unconscionable to help the industry at all by using the tech, which is a product of that industry.
api 6 hours ago||
It’s far less shitty than the social media industry (except where it overlaps) but that’s IMO.

Still kind of shitty. But if you really hate it use open models hosted commodity.

latent-9 8 hours ago||
i try kimi k3 for build ascii art ant calligram and than amazing result
m00dy 19 hours ago||
There’s going to be a lot of competition around this model. Let’s see how low AI providers are willing to push prices.
nickthegreek 8 hours ago||
They cant push it too low. The license agreement it is released under wont allow it.

> If the Licensee or any of its affiliates operates a Model as a Service business, and the aggregate revenue of the Licensee and its affiliates exceeds 20 million US dollars (or the equivalent in other currencies) in total over any consecutive 12 months, the Licensee must enter into a separate agreement with Moonshot AI before using the Software or its derivative works for any commercial purpose.

HDBaseT 3 hours ago||
Yeah, this license is a lot different to Kimi K2.7 Code or Kimi 2.6, previously it stated:

"Our only modification part is that, if the Software (or any derivative works thereof) is used for any of your commercial products or services that have more than 100 million monthly active users, or more than 20 million US dollars (or equivalent in other currencies) in monthly revenue, you shall prominently display "Kimi K2.7 Code" on the user interface of such product or service."

Although as you mentioned, now there is a revenue cap before custom agreements must take place. Currently there is 7 providers on OpenRouter for Kimi K3 and they have all the exact same price unfortunately.

torginus 18 hours ago|||
I think the results might be underwhelming - AI providers need to turn a profit and can't subsidize, and they're working off of the commodity hardware everyone does.

I wouldn't be surprised if they started offering potentiall bad quantizations with much reduced capability at lower prices (without telling the users, of course)

user43928 17 hours ago||
I would be surprised, considering that OpenRouter requires disclosing the quantization and shows automatic benchmarks to compare between providers for the same model.
m00dy 16 hours ago||
The latter is a joke
user43928 16 hours ago||
I saw that it runs GPQA Diamond and TAU-Bench Airline and shows the results over a 32 day rolling average.

Other than that they track Tool call error rate and Structured output error rate.

I only discovered this today, and it seems like a good idea. What are the problems in practice?

Iolaum 18 hours ago||
As long as they are transparent about what quant they serve the model and any other optimization they do that also affects performance of inferred tokens.
minimaxir 18 hours ago||
...does Hugging Face have enough bandwidth to let people download en masse however much file size a 2 trillion parameters model is?
cpburns2009 11 hours ago|
I have a feeling the number of people downloading 2t parameter models is significantly smaller than the 1-100b models.
qainsights 11 hours ago||
Seeing 404
taf2 11 hours ago||
it's a 404 link now
hellajack3d 18 hours ago||
So... Now we give huggingface the hug of death - right? ;)
seydor 16 hours ago||
It's thankful that openAI or anthropic haven't IPOed yet. Or terrible for some
dxxvi 16 hours ago|
Now I hope that nvidia will host it for free :-)
nateb2022 13 hours ago|
if it's anything like the speeds Nvidia Nim puts Deepseek at, it'll probably be 10 tok/s or lower and timeout frequently
More comments...