Training, testing & serving AI models.
Joined August 2026
- Tweets880
- Following337
- Followers1.1K
- Likes556
Pinned Tweet
Gemini 4 Pro (in arena) vs Claude Opus 5.5 on a 3D floatplane physics simulation
>Gemini 4 pro is completely outperforms Claude opus 5.5 here
> When it officially drops, it’s going to raise the bar completely.
Kimi K3.1 Pro vs. Gemini 4 Carbon (Arena Checkpoint) vs. Gemini 4 Argon vs. GPT-6 Astra
>Kimi K3.1 is giving frontier models some real competition and cooked Astra
>By the way, I am not sponsored by Kimi.
Testing Kimi K3.1 Pro on procedural web animation using Three.js.
>cyclist riding a bicycle
>As of now I am getting good results from Kimi K3.1
Kimi K3.1 Pro is definitely a highly capable model,but forcing influencers to hide paid partnerships and pre-approve every post just spoiled the launch.
That's how Kimi AI are promoting their new model Kimi K3.1 pro and I know all the influencers working with them and none of them are honest,I have posted about it myself but I didn't join this and never used their templates that they provide.
PROPAGANDA, I have more to share it's f**cked up I'm gonna have follow up for this
Kimi K3.1 Pro: 55k tokens
Gemini 4 Argon: 62k
Claude Opus 5.5: 119k
Same Intelligence Index tasks ,Kimi does it with less than half the tokens of Opus. Efficiency gap is real. ⚡
Artificial Analysis index benchmark
Claude Opus 5.5 : 57.60🥇
Claude Sonnet 5.5 : 56🥈
Muse Wonderfull : 55.80🥉
Kimi K3.1 Pro : 55.10💀
>open weights by Nov 4 ,will lead open-weights for long time
Kimi K3.1 pro vs. Opus 5.5 vs. GPT-6 Astra single html file three.js test.
>Kimi K3.1 cooked Astra
>Kimi came out of nowhere and is now competing at the frontier.
Kimi K3.1 Pro going toe-to-toe with Opus 5.5 and GPT-6 Astra,topping SWE Marathon, ZeroBench, and Humanity’s Last Exam.
First test with Kimi K3.1 Pro: a galloping horse in Three.js.
> Kimi, you beauty! It looks really good.
>Totally different output from what I got from Arena yesterday.
Kimi K3.1 is officially out now after being featured in the arena under the name 'kibitz'.
>Kimi AI with K3.1 pro | Built for Coding & Agentic Work
Claude Sonnet 5.5 now sending every request straight to Fable 5.5.
>sonnet 5.5 routing to Fable 5.5
>I got a few outputs from Sonnet 5.5, but they're definitely not Sonnet 5.5's.
Gemini 4 Argon 2nd checkpoint in Arena vs. Opus 5.5 vs. Kimi K3.1 in Arena
>Carbon >~ Opus 5.5
>Kimi K3.1 is good in its own league
These are the models that are going live in Arena:
>Gemini 4 Carbon under the name gemini 4 Argon
>Kimi K3.1 under the name Kibitz
>Qwen 4 under the name Kalmia
Gemini 4 Carbon checkpoint in Arena vs. Opus 5.5 vs. GPT-6 Astra
>Clearly, Carbon outperforms Opus 5.5 on this Three.js test.
>Carbon definitely takes the top stop when it is officially released.
Gemini 4 Carbon checkpoint in the Arena under the name Gemini 4 Argon:
>A freestyle swimmer doing the front crawl using Three.js (single HTML file)
I got few exceptional outputs from Gemini 4 Carbon checkpoint in Arena under the name ‘gemini 4 argon’
>Matching Opus 5.5 closely,even edging it out in a few cases.(as of now)
Gemini 4 Carbon checkpoint in Arena vs. Argon vs. GPT-6 Astra
>Carbon is already better than Argon
>GPT-6 Astra is starting to look outdated.