Meet Kimi K3

god knows how hard to keep my mouth shut!!!

Kimi entering the market :

@nicdunz

>Scott Alexander makes a post about how Chinese AI is 6-12 months behind >Chinese AI catches up less than a week later Well that aged poorly lmao

Dario after meeting K3

This is no longer ‘catching up’ it’s competing.

So are we all paying for Dario's therapy or no?

Fable 5 and GPT 5.6 watching

Quick summary below, but in short it's the new and current SoTA Opensource model Great job, Kimi bros!!!

fable 5 and gpt 5.6 watching

the second deepseek r1 moment

K3 make Mythos 6 and MAKE NO MISTAKE .

OH MY GOD, GAME DEV IS GOING TO BE SO MUCH BETTER IN THE NEXT COUPLE OF MONTHS!

MOG THEMMMM

Really good

Kimi K3 surpassed GPT 5.6 Sol on the Vals Index! Check it out

this is amazing, huge congrats and thanks kimi team (looking forward for the tech report as always 🫡)

Help I think something is wrong with the chat function??

K3 is here 😍 @opencode please do the needful (3x limit would be muahh)

SENDDDD ITTTTTTTTTTT

Dario and Sam are cooked

It's official launched!

they distilled even Anthropic's marketing lol

Wake up, babe. New model just dropped.

Congratulation!

dope video 🔥

Deepseek moment nah it’s kimi k3 moment! This is historical

Can somebody familiar with the benchmark explain to me like I'm 12. Is the result just 3-5% better or is this score non-linear (so its substantially better than 3-5%?). Because top-1 result doesn't look tremendously more impressive than the bottom of the chart in %.

Full comparison with Fable 5, here