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

Kimi K3: Open Frontier Intelligence

kimi.com

Post reply on HN
Print view View original

891–900 of 1001 posts

Re: Kimi K3: Open Frontier Intelligence

#891

Post by daishi55 »

mrngld wrote: post #856
oefrha wrote: post #783

Earlier quoted context omitted.

Apparently open source models is an officially stated national strategy now, which is somewhat reassuring that we'll probably keep getting these for a while. https://www.reuters.com/world/asia-pacific/chinas-xi-promote...

Until it's no longer officially stated strategy, and something else benefits the CCP. See rare earths policy. Changes constantly, to whatever maximizes their leverage. The aspect that you can count on is that China sees AI as a long term strategic necessity, so they will support domestic players for an extremely long time. But unless you're a Chinese lab that shouldn't necessarily be a comfort to anyone else.

Unlike every other country on the planet which use their leverage to benefit all nations equally.

Re: Kimi K3: Open Frontier Intelligence

#893

Post by pegasus »

orbital-decay wrote: post #749
katzenversteher wrote: post #718

Earlier quoted context omitted.

Hey Simon, I noticed one thing all LLMs are currently pretty bad at and maybe we could create a benchmark from it. Let an LLM play the role of a dungeon master and tell it to strictly stay in the script/story and only allow realistic player actions. You will notice that they are easily brought off track. E.g. - Tell the LLM that you as a player noticed a strange glow in an NPCs eyes -> the NPC becomes an enemy. - In…

This is a known and solved problem. Such a test is pointless for a general-purpose model, because like most people you're using multiturn chats in a naive way, fighting the default finetuning that is done intentionally. 1. You're sending your in-character inputs to an instruction-tuned model under the user role, in a multiturn chat. It's biased to treat these inputs as instructions and this behavior will show itself…

Prefill isn't supported in the major commercial models anymore. Probably in order to thwart jailbreak attacks.

Re: Kimi K3: Open Frontier Intelligence

#894

Post by mattstir »

abalashov wrote: post #863

I switched to exclusively Chinese models, mostly Kimi, many months ago. I'll still ask Claude questions that require ambitious real-time web search / worldly knowledge, but for just about anything else, the Chinese models have been so good that I haven't looked back.

I think that those two statements are somewhat contradictory of one another. But in any case, I'm curious about your decision to still use Claude for some questions. Do you find the Chinese models have less worldly knowledge, and if so in what categories? Genuinely curious, I haven't had a chance to try them out very much.

Re: Kimi K3: Open Frontier Intelligence

#895

Post by logicallee »

InsideOutSanta wrote: post #173

On the first try, Kimi K3 just found the source of a bug that Fable 5 hasn't been able to pinpoint in multiple attempts. It's just one anecdote, and I haven't used K3 much yet, but so far it's looking extremely promising.

>On the first try, Kimi K3 just found the source of a bug that Fable 5 hasn't been able to pinpoint in multiple attempts

Very interesting, thanks for sharing! Could you give some details about what kind of software (language or environment) and what kind of bug it was? Was it a single-file bug, like could it fit in one context like a chat window, or were you using an agentic version (Kimi Code) that looked through multiple files and then found a bug that manifested through complex interactions of multiple systems/files?

Re: Kimi K3: Open Frontier Intelligence

#896

Post by orbital-decay »

pegasus wrote: post #893
orbital-decay wrote: post #749

Earlier quoted context omitted.

This is a known and solved problem. Such a test is pointless for a general-purpose model, because like most people you're using multiturn chats in a naive way, fighting the default finetuning that is done intentionally. 1. You're sending your in-character inputs to an instruction-tuned model under the user role, in a multiturn chat. It's biased to treat these inputs as instructions and this behavior will show itself…

Prefill isn't supported in the major commercial models anymore. Probably in order to thwart jailbreak attacks.

Only on GPT (never was) and more recently Claude, and it still can be simulated with structured outputs for the purpose above.

Even if it's not supported somewhere (e.g. z.ai API which isn't mature enough and has neither assistant prefills nor actual structured outputs), it's still better and more seamless than using the default user/assistant scaffolding for role alternation.

Re: Kimi K3: Open Frontier Intelligence

#897

Post by fc417fc802 »

katzenversteher wrote: post #736
pitched wrote: post #725

Earlier quoted context omitted.

An approach I like to help solving this is antagonistic or review agents. The first agent decides that eye glows turn NPCs into enemies, the second agent is fully dedicated to deciding if that is valid. If the review fails, it leaves notes and the original agent tries again.

So far I only tried it with a single LLM in the dungeon master role. Your approach sounds promising (and I will definitly try it) but also a bit like a complicated workaround. What I mean: In games with humans the dungeon master is usually one person, not a whole council ;)

But any competent human is also playing multiple roles mentally, mentally asking an entire series of questions about any new information. Discreet rounds of review emulate that. Write down the human process as a flow chart and then each interior node in the chart becomes a discreet review step with its own prompt.

Re: Kimi K3: Open Frontier Intelligence

#898

Post by mordymoop »

softwaredoug wrote: post #478

So Chinese labs are driving essentially towards commodotized intelligence. Even if its a few months behind the US. Is this a classic 'commoditize my compliment' situation? They want to sell the hardware and infrastructure behind AI and make the software part not the value driver / moat? I can see it. But also even two Chinese labs sinking 100s of millions USD into training isn't exactly commoditization. It's still a…

To your last point, it can’t possibly be sustainable, so it reads to me more as a short term FUD attack on American dominance in this space. It might have cost them half a billion dollars to train this model, and they’re going to make nothing off it. How many more times can they afford to do that? It’s going to get more expensive to train AI going forward, not less.

I also have a suspicion that the benchmark numbers are not real.

Re: Kimi K3: Open Frontier Intelligence

#899

Post by simonw »

RA_Fisher wrote: post #888
simonw wrote: post #111

Pelican: https://tools.simonwillison.net/markdown-svg-renderer#url=ht... - rendered via the OpenRouter API: https://openrouter.ai/moonshotai/kimi-k3 95 input, 16,658 output = 25 cents! https://www.llm-prices.com/#it=95&ot=16658&ic=3&oc=15 (13,241 of those were reasoning tokens.) I think that's the most expensive pelican I've rendered through a Chinese model so far.

We need a new Pelican, bc earlier models were disadvantaged relative to later models. It's not a very useful comparison anymore.

I wrote about what it's useful for (not much these days) yesterday: https://simonwillison.net/2026/Jul/16/kimi-k3/#what-can-we-l...

Re: Kimi K3: Open Frontier Intelligence

#900

Post by jp0001 »

I really don't see how the SaaS models will be profitable if this continues, all the money will be in providing consumers with hardware that can run these locally - with the ability to modify the weights and do your own ablation.
Post reply on HN