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

Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

kimi.com

Post reply on HN
Print view View original

161–170 of 251 posts

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#161

Post by storystarling »

jumploops wrote: post #4

> For complex tasks, Kimi K2.5 can self-direct an agent swarm with up to 100 sub-agents, executing parallel workflows across up to 1,500 tool calls. > K2.5 Agent Swarm improves performance on complex tasks through parallel, specialized execution [..] leads to an 80% reduction in end-to-end runtime Not just RL on tool calling, but RL on agent orchestration, neat!

1,500 tool calls per task sounds like a nightmare for unit economics though. I've been optimizing my own agent workflows and even a few dozen steps makes it hard to keep margins positive, so I'm not sure how this is viable for anyone not burning VC cash.

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#162

Post by HNisCIS »

esafak wrote: post #144
Alifatisk wrote: post #53

Have you all noted that the latest releases (Qwen3 max thinking, now Kimi k2.5) from Chinese companies are benching against Claude opus now and not Sonnet? They are truly catching up, almost at the same pace?

They are, in benchmarks. In practice Anthropic's models are ahead of where their benchmarks suggest.

Bear in mind that lead may be, in large part, from the tooling rather than the model

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#163

Post by zozbot234 »

storystarling wrote: post #161
jumploops wrote: post #4

> For complex tasks, Kimi K2.5 can self-direct an agent swarm with up to 100 sub-agents, executing parallel workflows across up to 1,500 tool calls. > K2.5 Agent Swarm improves performance on complex tasks through parallel, specialized execution [..] leads to an 80% reduction in end-to-end runtime Not just RL on tool calling, but RL on agent orchestration, neat!

1,500 tool calls per task sounds like a nightmare for unit economics though. I've been optimizing my own agent workflows and even a few dozen steps makes it hard to keep margins positive, so I'm not sure how this is viable for anyone not burning VC cash.

"tool call" is just a reference to any elementary interaction with the outside system. It's not calling third-party APIs or anything like that.

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#164

Post by two_tasty »

whizzter wrote: post #92
motoboi wrote: post #80

Earlier quoted context omitted.

What amazes me is why would someone spend millions to train this model and give it away for free. What is the business here?

Chinese state that maybe sees open collaboration as the way to nullify any US lead in the field, concurrently if the next "search-winner" is built upon their model the Chinese worldview that Taiwan belongs to China and Tiamen Square massacre never happened. Also their license says that if you have a big product you need to promote them, remember how Google "gave away" site searche widgets and that was perhaps one of…

I love how Tiananmen square is always brought up as some unique and tragic example of disinformation that could never occur in the west, as though western governments don't do the exact same thing with our worldview. Your veneer of cynicism scarcely hides the structure of naivety behind.

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#165

Post by Alifatisk »

WarmWash wrote: post #135
Alifatisk wrote: post #53

Have you all noted that the latest releases (Qwen3 max thinking, now Kimi k2.5) from Chinese companies are benching against Claude opus now and not Sonnet? They are truly catching up, almost at the same pace?

They distill the major western models, so anytime a new SOTA model drops, you can expect the Chinese labs to update their models within a few months.

Yes, they do distill. But just saying all they do is distill is not correct and actually kind of unfair. These Chinese labs have done lots of research in this field and publish it to the public, some of not majority contribute with open-weight models making a future of local llm possible! Deepseek, Moonshot, Minimax, Z.a, Alibabai (Qwen).

They are not just leeching here, they took this innovation, refined it and improved it further. This is what the Chinese is good at.

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#166

Post by nostrebored »

Jackson__ wrote: post #28

As your local vision nut, their claims about "SOTA" vision are absolutely BS in my tests. Sure it's SOTA at standard vision benchmarks. But on tasks that require proper image understanding, see for example BabyVision[0] it appears very much lacking compared to Gemini 3 Pro. [0] https://arxiv.org/html/2601.06521v1

Gemini remains the only usable vision fm :(

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#167

Post by symisc_devel »

throwaw12 wrote: post #67

Congratulations, great work Kimi team. Why is that Claude still at the top in coding, are they heavily focused on training for coding or is it their general training is so good that it performs well in coding? Someone please beat the Opus 4.5 in coding, I want to replace it.

Gemini 3 pro is way better than Opus especially for large codebases.

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#168

Post by WhitneyLand »

zozbot234 wrote: post #160
WhitneyLand wrote: post #129

Earlier quoted context omitted.

Its often pointed out in the first sentence of a comment how a model can be run at home, then (maybe) towards the end of the comment it’s mentioned how it’s quantized. Back when 4k movies needed expensive hardware, no one was saying they could play 4k on a home system, then later mentioning they actually scaled down the resolution to make it possible. The degree of quality loss is not often characterized. Which makes…

> ...Back when 4k movies needed expensive hardware, no one was saying they could play 4k on a home system, then later mentioning they actually scaled down the resolution to make it possible. ... int4 quantization is the original release in this case; it's not been quantized after the fact. It's a bit of a nuisance when running on hardware that doesn't natively support the format (might waste some fraction of memory t…

Good point thanks for the clarification.

The broader point remains though which is, “you can run this model as home…” when actually the caveats are potentially substantial.

It would be so incredibly slow…

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#169

Post by WhitneyLand »

FuckButtons wrote: post #149
WhitneyLand wrote: post #129

Earlier quoted context omitted.

Its often pointed out in the first sentence of a comment how a model can be run at home, then (maybe) towards the end of the comment it’s mentioned how it’s quantized. Back when 4k movies needed expensive hardware, no one was saying they could play 4k on a home system, then later mentioning they actually scaled down the resolution to make it possible. The degree of quality loss is not often characterized. Which makes…

From my own usage, the former is almost always better than the latter. Because it’s less like a lobotomy and more like a hangover, though I have run some quantized models that seem still drunk. Any model that I can run in 128 gb in full precision is far inferior to the models that I can just barely get to run after reap + quantization for actually useful work. I also read a paper a while back about improvements to mo…

Interesting.

If this were the case however, why would labs go through the trouble of distilling their smaller models rather than releasing quantized versions of the flagships?

Re: Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

#170

Post by igneo676 »

two_tasty wrote: post #164
whizzter wrote: post #92

Earlier quoted context omitted.

Chinese state that maybe sees open collaboration as the way to nullify any US lead in the field, concurrently if the next "search-winner" is built upon their model the Chinese worldview that Taiwan belongs to China and Tiamen Square massacre never happened. Also their license says that if you have a big product you need to promote them, remember how Google "gave away" site searche widgets and that was perhaps one of…

I love how Tiananmen square is always brought up as some unique and tragic example of disinformation that could never occur in the west, as though western governments don't do the exact same thing with our worldview. Your veneer of cynicism scarcely hides the structure of naivety behind.

The difference is that, in the west, there's an acceptable counter narrative. I can tell you that Ruby Ridge and Waco never should've happened and were examples of government overreach and massacre of it's own citizens. Or

You can't with Tiananmen square in China

Post reply on HN