Zheng | Vibe Working
@zheng401
Followers
734
Following
51
Media
49
Statuses
140
vibe living with AGI @Kimi fueled by caffeine, code and dream！
Joined March 2009
From 10 tok/s ➡️ up to 100 tok/s 🚀.kimi-k2-turbo-preview just got another massive speed boost ⚡️.Normal: 60 tok/s.Off-peak: 100 tok/s.🎥 The speed jump is crystal clear in the video!
1
0
3
Manus->lovart->Opus Clip?.
In the AI world, consensus can flip in 3 months. But one golden rule remains: the new world rewards speed. Shipping Genspark AI Designer this week!
1
0
1
This is something Generative AI can never compete with.
I have never seen anything like this, China's tribute to Steph Curry last night is amazing 🇨🇳 🏀
0
0
5
Unitree won the 400m race in the humanoid robot sports games! This year's tickets are already incredibly popular—next year will probably be even hotter.
1
0
5
This video helped Insta360 Gain 2bn USD in market cap today!.
Witness the unveiling of Antigravity A1, the 8K 360 drone. And this is only the first half of the journey. #Antigravity #AntigravityA1 #Insta360 #AntigravityDrone #360Drone
1
0
1
OpenRouter's API usage has reached new heights, emerges as the biggest winner in the intense LLM competition over the past month. 🚀 #AI #LLM #OpenRouter
0
1
3
It seems that few people are talking about GPT6 after the launch of GPT5. the 1st time since the launch of chatgpt.
2
0
6
DJI is launching a 360° camera, taking aim at Insta360's core business. Meanwhile, Insta360 is rumored to be working on a drone. The competition just got even fiercer! #DJI #Insta360 #TechBattle
1
0
4
the US AI action plan, encouraging something mostly made in China
0
2
4
asked Kimi K2: search info and compare Kimi K2 and Qwen3-Coder-480B-A35B-Instruct, compare these 2 models and give me a conclusion about which one is better, show highlights in one page html，Kimi is very candid and doesn't hesitate to praise the strengths of its peers.
0
1
13
"Benchmark" from real world jobs!.
I'd like to point out that for the real world tasks (not benchmarks), Kimi K2 outperforms Gemini. This is telemetry across all @cline users, showing diff edit failure rate. Notice how Kimi has about a 6% failure rate, which is significantly better than Gemini's ~ 10% error
0
1
7
