Advertisement
Xinmei Shen
Reporter, Technology
Xinmei Shen joined the Post in 2017 and is a technology reporter. She covers content, entertainment, social media and internet culture. Previously, she was with the Post’s tech news site, Abacus. Before that, she was a reporting intern at The Information whilst studying at the University of Hong Kong.
Xinmei Shen joined the Post in 2017 and is a technology reporter. She covers content, entertainment, social media and internet culture. Previously, she was with the Post’s tech news site, Abacus. Before that, she was a reporting intern at The Information whilst studying at the University of Hong Kong.
Areas of Expertise:
China technology coverage, video productionLanguages Spoken:
English, Mandarin
Kimi K3’s overall score of 32.2 per cent compares with top, unnamed US models that average 76.2 per cent.
Researchers point to the lack of evidence for any distillation claims, while some highlight Kimi K3’s technical innovations.
The incident fuels growing debate over the rapid advancement of autonomous AI systems capable of exploiting software vulnerabilities.
New model from Moonshot AI is ‘extremely close’ to OpenAI’s flagship after ‘very big jump’ in capabilities, researcher says.
Advertisement
Analysts say Beijing-based start-up Moonshot’s 2.8 trillion-parameter large language model could narrow the gap with the US within ‘weeks’.
Chinese start-up pauses Kimi K3 subscriptions due to surging demand and compute shortages, highlighting the country’s AI chip challenges.
Project is expected to churn out 60 trillion tokens a year, driving the country’s AI clean-power ambitions forward.
The speech marks the Chinese president’s first-ever keynote address at the nation’s premier artificial intelligence summit.
A growing number of tech firms are pivoting to lightweight models designed to run entirely on smartphones and laptops.
A sharp narrowing of the performance gap is prompting some developers to shift from costly American giants to budget-friendly Chinese alternatives.
As the city looks to challenge the US as a global crypto hub, Baillie Gifford says the SFC has approved its blockchain-based tokenised fund.
AI labs Yoolee and InfiX.ai say they aim to rival US start-up with industry-specific AI solutions, reduced costs and enhanced privacy.
Yan Junjie said the company remains committed to its long-term mission despite ‘external noise’ and market fluctuations.
As cutting-edge models like Zhipu AI’s GLM-5.2 rapidly close the performance gap with frontier models, policy analysts warn a regulatory shift may be imminent.
Alibaba vice-president Qi Xiaoning says global tech firms continue to build a unified, open-standard ecosystem amid rising geopolitical barriers.
The Chinese tech giant has listed Claude Code as ‘high-risk’ software, after discovering Anthropic had tracked Chinese users using hidden code.
Firm also known as Z.ai rolls out promotions to lure developers as US rival Anthropic responds to claim it covertly tracked Chinese users.
LongCat-2.0 boasts 1.6 trillion parameters and a million-token context window, on par with DeepSeek’s latest flagship model.
UC Berkeley’s RDI centre earlier this month introduced Agents’ Last Exam, a new benchmark that tests how well AI agents perform.
While other Chinese releases have made significant gains, GLM-5.2 is the first model to rank in the top three globally on a major benchmark.
Anthropic’s Mythos has become a ‘cyber nuclear weapon’ that China cannot afford to leave in US hands alone, says 360 founder Zhou Hongyi.
US$1.15 billion investment underscores advanced packaging’s role in boosting China’s chipmaking capabilities amid rapid AI development.
The aggressive price cut comes amid intense rivalry between US and Chinese artificial intelligence firms.
Anthropic’s global suspension of Fable 5 has sparked concerns about the reliability of US AI infrastructure.
Jack Ma and executives plant rice seedlings in a symbolic retreat signalling resilience as Alibaba doubles down on AI.
Global No 2 ranking for Zhipu’s breakthrough AI model triggers Musk exchange over whether China can close gap with US sooner than expected.
The move aims to support listings of ‘high-quality’ AI firms that have yet to form ‘a certain scale of revenue’.
A growing number of Chinese AI labs are experimenting with shifting earlier model training phases onto domestic chips.
Restrictions on access to the powerful model, ‘targeted at Chinese AI labs’, are sparking debate about safety, innovation and open research.
