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

Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

fireworks.ai

Post reply on HN
Print view View original

331–340 of 491 posts

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#331

Post by Slartie »

epolanski wrote: post #330
Slartie wrote: post #327

Earlier quoted context omitted.

I'd guess the slowness is mostly due to there currently being only one provider, Moonshot AI. And they are overwhelmed with demand. Let's judge the speed of the model when its weights are released and every inference provider on the planet offers it, so demand can spread out a bit. It's the same topic with token budget comparisons and subscription pricing - don't people understand that this doesn't really matter for…

Even if hardware capacity increases, it seems clear it uses way more tokens, so I don't expect parity with other competitors on that front. On the other hand I expect K3 future refinements to be massive and more efficient.

The speed at which tokens are crunched, even on the same hardware, differs between models as well. Using more tokens is only a problem if they are processed at the same speed as with a comparison model.

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#332

Post by cdud3 »

bmitc wrote: post #325
nxtfari wrote: post #253

If you haven’t been really running and testing these models yourself, they are all benchmaxxed. No matter how close they score to frontier on whatever metric, they always fall apart in real world tasks and their token efficiency is ridiculously bad. Fireworks has incredible incentive to make this claim in a headline, because Fireworks hosting K3 for you is pure profit for them, unlike when they host closed source mod…

Why is token efficiency a concern with free models?

Because that's the only argument left after Kimi beats Fabel in results, price and autonomy.

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#333

Post by Maxious »

scilro wrote: post #229
deminature wrote: post #221

Earlier quoted context omitted.

Fireworks (the author of OP's article) is a western provider based in San Mateo, California

Fireworks isn't serving Kimi K3 yet. Presumably, they ran this benchmark against the Moonshot API. All of the Western providers with sufficient capacity will be able to make it available when the weights are released Monday.

Fireworks has been working on porting the Kimi Delta Attention (KDA) hybrid linear attention mechanism to their hosting infrastructure https://x.com/FireworksAI_HQ/status/2079776331609584005

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#334

Post by swiftcoder »

johnhess wrote: post #73

Was this an out of sample test of the router or was it trained on these specific use cases/eval suites?

Neither. They ran all the cases using both models, picked the winning result for each one, and then said "if you had a router that guessed with 100% accuracy, here's what it would have picked"

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#335

Post by BlobberSnobber »

nonadhocproblem wrote: post #303
throwaway277432 wrote: post #297

Earlier quoted context omitted.

Claude (Opus 4.8) recently told me: > I’d ask you to drop the abuse; (...) if it continues I’ll end the conversation. After I'd used a couple of expletives. And yes it will emit a token. This is truly dystopian. It is NOT a person. What a response. I still cant believe it.

So you're distraught at losing the ability to abuse digital minds? Excellent, I'm glad Anthropic introduced this.

There’s no way you actually believe these word-predictors are actually thinking, right?

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#336

Post by olmo23 »

throwaway277432 wrote: post #297
4b11b4 wrote: post #218

Earlier quoted context omitted.

I strictly prefer when models ignore any human quirks in my responses. Claude trying to be your friend, saying LOL to your jokes is ridiculous and frankly, harmful

Claude (Opus 4.8) recently told me: > I’d ask you to drop the abuse; (...) if it continues I’ll end the conversation. After I'd used a couple of expletives. And yes it will emit a token. This is truly dystopian. It is NOT a person. What a response. I still cant believe it.

It's because some people within Anthropic refuse to rule out the possibility that LLMs have the ability to suffer. If you ask Claude, he'll tell you all about it.

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#337

Post by Maxious »

claytonjy wrote: post #291
ralusek wrote: post #36

Earlier quoted context omitted.

One benefit of an open source one is that you can, as a large corporation, run it "locally" within your own data center. Even fine tune it.

How big is this market, self-hosting a model that requires 64 GPUs, H100 or better, with good interconnects between nodes? I suspect the overlap of those that can afford it, and those that have the talent to manage it, is a fairly thin slice of the Venn diagram. Even the large corps are gonna be getting it from the inference vendors, or more likely Bedrock and friends.

Dell will sell you a PowerEdge XE7740/XE7745 "AI Factory" with 32 H200s https://infohub.delltechnologies.com/en-au/t/dell-ai-factory...

They "booked $24 billion in AI server orders this quarter as its customer base broadened to more than 5,000" https://finance.yahoo.com/markets/stocks/articles/dells-ai-f...

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#338

Post by fractalf »

4b11b4 wrote: post #218
abdullahkhalids wrote: post #212

I will accept a 5% drop in benchmarks for a model that talks to me like a human.

I strictly prefer when models ignore any human quirks in my responses. Claude trying to be your friend, saying LOL to your jokes is ridiculous and frankly, harmful

You tell jokes to your model? :D

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#339

Post by vorticalbox »

nxtfari wrote: post #253

If you haven’t been really running and testing these models yourself, they are all benchmaxxed. No matter how close they score to frontier on whatever metric, they always fall apart in real world tasks and their token efficiency is ridiculously bad. Fireworks has incredible incentive to make this claim in a headline, because Fireworks hosting K3 for you is pure profit for them, unlike when they host closed source mod…

I encountered the same issue. Kimi 2.7 looked impressive on paper, but in practice, the code was so riddled with errors that I ended up using GPT-5.5 to fix it.

Re: Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA

#340

Post by throwaway277432 »

nonadhocproblem wrote: post #303
throwaway277432 wrote: post #297

Earlier quoted context omitted.

Claude (Opus 4.8) recently told me: > I’d ask you to drop the abuse; (...) if it continues I’ll end the conversation. After I'd used a couple of expletives. And yes it will emit a token. This is truly dystopian. It is NOT a person. What a response. I still cant believe it.

So you're distraught at losing the ability to abuse digital minds? Excellent, I'm glad Anthropic introduced this.

Yes, in the same way I like to kill the enemies in DOOM.

It's matrix multiplication. Absurd.

Post reply on HN