Kimi K3 absolutely mogs every open source model this is on par , head on head all case i tried in coding with fable , i feel weird to write , but sometimes it output better quality than fable too I don't know how to comprehend this , i... have lots of test and comparison 😼show more

Average Google Deepmind Researcher:

I’m excited to run it on my 4gb laptop

impossible lol

It will work on 1mil context 😂

Kimi K3 has genuinely surprised me too, wasn't expecting it to hit this hard 👀

fable still better looking but this is insane. cant wait to try it

This model doesn't need Trump's approval, right?

I’m desperate for this one. Any idea of the release date?

@VictorTaelin big if true

Can you try this?

Awesome, the only surprise is how fast this happened.

People sleep on Kimi because of the price. Neither the most expensive nor cheapest, middling speed.

Just because it’s close to fable on threejs visualizations doesn’t mean it’s close to fable along most other axes, does it?

i want test it

Does that mean they're now ahead or simply that gpt 6 is at our doorstep?

opensourcemogging

ohh so kia, the car brand means killed in action, okay

damn bro

有竞争才是好事。 模型时代可能不会是谁永远坐庄，而是谁能在不同场景把成本、速度、生态和性能做到最佳平衡。Kimi 这波如果能稳定复现，确实值得关注。

刚看到你上一个帖子就说不要消息k3炒作，😅

kimi k3 beating fable on demos is real but repo scale tests decide

HOLY Didn’t Kimi K2.7 release a few weeks ago??

Its weird because you say this, i open the the video, and the characters heads are not even attached to the shoulders lol. Not saying its impressive, but its certainly not a 1 shot monster.

🤡

wow, i will try some complex game prompts with K3 once it's available. currently GLM-5.2 is best, maintaining a testing thread myself.

I've been watching these model jumps closely. My take is hands on testing keeps becoming more important than leaderboard assumptions.

Open weights quietly cooking while everyone watches the closed model leaderboard

Una pregunta de tu opinión tu crees que Kimi k3 es una destilación de Fable 5 o no

真的假的啊，这些测试工程能公开么？ 还是老哥你收钱办事儿的。

Link to this game?

It is totally valid to trust your results. Kimi K3's coding and reasoning leaps show that top open-source models are officially matching—and sometimes beating—the best closed-source options.

Also, where do I get K3?!

Nice

Cyber abilities 👽👽