Artificial Intelligence thread

H

HighGround

Senior Member
Registered Member
mossen said:
Having tried the new Kimi K2.5 and Qwen Max-Thinking models I can only say... the gap between Qwen and Kimi is getting larger, not smaller. Kimi is simply in another league. Qwen not only did poorly, it straight-up hallucinated on several of my questions.

K2 was already a good model but it is now clearly great with the 2.5 update. The reasoning performance is on par with Opus 4.5 and GPT5.2-high in my testing.

Of all the major Chinese labs, Moonshot is probably most deserving of being flooded with GPUs and investment. Too bad geopolitics will prevent it. The talent density of the team is astounding.
Click to expand...
The worth of an AI isn't tied to their hallucination rate is all I'll say. Qwen's stack is much more important than Kimi.
 
S

Some1Guy

New Member
Registered Member

Chinese AI goes next level in geometry at a top US maths Olympiad​

Please, Log in or Register to view URLs content!
A Chinese
Please, Log in or Register to view URLs content!
system has outperformed its US competitors in solving geometry problems at the International Mathematical Olympiad (IMO) level, taking less than half the time and using simpler computational resources, according to its developers.
Click to expand...
Unlike existing models confined to problem-solving, the Chinese system can also generate mathematical problems – three of which appeared in a Chinese national team qualifying exam and a top Olympiad in the United States in 2024.
Click to expand...
“We present TongGeometry, a neuro-symbolic system that discovers, proposes and proves IMO-level geometry problems through principled tree search,” researchers from the Beijing Institute for General Artificial Intelligence and
Please, Log in or Register to view URLs content!
wrote in the peer-reviewed journal Nature Machine Intelligence on Monday.
Click to expand...
The developers of TongGeometry said the system “functioned more like a coach who both designs training problems and guides solution strategies, rather than merely a student who solves given problems”.
Click to expand...
Drawing on 196 past Olympiad geometry problems, the system generated 6.7 billion geometry problems that required auxiliary constructions.
Click to expand...
 
Wrought

Wrought

Captain
Registered Member
Reuters is reporting that Deepseek and others have been granted conditional approval to purchase H200s at scale.

SINGAPORE, Jan 30 (Reuters) - China has given its top AI startup DeepSeek approval to buy Nvidia's
Please, Log in or Register to view URLs content!
H200 artificial intelligence chips with regulatory conditions that are still being finalised, two people familiar with the matter told Reuters. Reuters
Please, Log in or Register to view URLs content!
citing sources, that ByteDance, Alibaba
Please, Log in or Register to view URLs content!
and Tencent
Please, Log in or Register to view URLs content!
had been given permission to purchase more than 400,000 H200 chips in total.

Nvidia CEO Jensen Huang told reporters in Taipei on Thursday that his company
Please, Log in or Register to view URLs content!
received such information. He added that he believed that China was still finalising the licence. Nvidia did not respond to a request for comment on DeepSeek's approval. China's industry and commerce ministries have granted approvals for all four companies, but have stipulated that they will impose conditions that are still being finalised, the sources said. These conditions are being decided by China's state planner, the National Development and Reform Commission (NDRC), according to one of the people.
Click to expand...

Please, Log in or Register to view URLs content!
 
You must log in or register to reply here.
Top