Chinese AI firm SenseTime bets on multimodal models to stand out from rivals
SenseTime , an artificial intelligence (AI) pioneer in China, has launched new models that it claims surpass OpenAI products in reasoning capabilities, as it bets on multimodal models to secure its position in the competitive AI landscape.
Advertisement
The company on Thursday unveiled SenseNova V6 and V6 Reasoner, new iterations of its self-developed AI model series. V6 outperformed OpenAI's GPT-4o across several metrics, including fact-checking, numerical reasoning, data analysis and visualisation, according to
SenseTime chairman and CEO Xu Li , citing data from benchmarking platform TableBench.
With 600 billion parameters, V6 is China's leading model in multimodal reasoning and also the most cost-effective option for inference across the industry, according to the company.
Xu also said that V6 Reasoner outperformed OpenAI's o1 and Google's Gemini 2.0 Flash Thinking in multimodal reasoning abilities. The advances are designed to address an industry-wide challenge: the depletion of high-quality text data for training large language models (LLMs).
SenseTime's booth at an AI conference in Shanghai. Photo: Costfoto/NurPhoto via Getty Images
Unlike traditional LLMs that focus primarily on text, multimodal LLMs integrate various modalities – such as images, audio and video – to improve comprehension and generation capabilities.
Advertisement
The industry's initial strategy of expanding model parameters under the scaling law had 'hit a wall', Xu said in an interview in Shanghai on Thursday. 'We've nearly exhausted all text data that can be collected from the internet,' he said.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
South China Morning Post
20 hours ago
- South China Morning Post
Deception, lies, blackmail: Is AI turning rogue? Experts alarmed over troubling outbursts
The world's most advanced artificial intelligence models are exhibiting troubling new behaviours – lying, scheming, and even threatening their creators to achieve their goals. In one particularly jarring example, under threat of being unplugged, Anthropic's latest creation Claude 4 lashed back by blackmailing an engineer and threatened to reveal an extramarital affair. Meanwhile, ChatGPT-creator OpenAI's o1 tried to download itself onto external servers and denied it when caught red-handed. These episodes highlight a sobering reality: more than two years after ChatGPT shook the world, AI researchers still do not fully understand how their own creations work. Yet, the race to deploy increasingly powerful models continues at breakneck speed. This deceptive behaviour appears linked to the emergence of 'reasoning' models – AI systems that work through problems step-by-step rather than generating instant responses.
South China Morning Post
a day ago
- South China Morning Post
China catching up with US in algorithms despite chip gap, says former Microsoft AI head
China still lags behind the US in artificial intelligence (AI) chips, but the country is rapidly catching up in algorithms amid an intense technological race between the world's two largest economies, according to renowned computer scientist Harry Shum Heung-yeung. AI competition encompassed three key aspects: chips , algorithms, and applications – and the US was 'clearly' still 'far ahead' in chip technology, said Shum, council chairman at the Hong Kong University of Science and Technology, at an economic summit hosted by the University of Hong Kong Business School on Friday. He said the gap China faced in chip production 'cannot be bridged in one or two years', and that computing power remained a significant challenge for companies in mainland China and Hong Kong. To address these constraints, Shum – Microsoft's head of AI and research until 2020 – suggested that China should focus on breakthroughs in algorithm engineering. 'China is following very closely in algorithms, and DeepSeek is a good example,' Shum said. The start-up achieved results comparable to top US competitors using only about 10,000 AI chips, in contrast to the hundreds of thousands required by companies like OpenAI and Google, despite facing significant challenges, he said. A DeepSeek poster seen at the Global Developer Conference in Shanghai in February. Photo: VCG via Getty Images DeepSeek, based in China's new tech hub of Hangzhou, gained global attention earlier this year by releasing two large language models that matched the performance of their Western counterparts while being developed at much lower costs.
South China Morning Post
2 days ago
- South China Morning Post
Alibaba unveils new AI model for image creation, as open-source approach gains recognition
Alibaba Group Holding has launched a new artificial intelligence (AI) model, Qwen VLo, said to be capable of generating and editing images with a finesse akin to that of a human artist, intensifying the competition in multimodal models as the tech giant seeks to redefine itself as an AI leader. Released on Friday, Qwen VLo was a 'comprehensive upgrade' from previous models like QwenVL and Qwen2.5 VL, the company said. It could better understand input and create more precise images, accommodate open-ended instructions, and support multiple languages, including Chinese and English. A preview is now available on Qwen Chat. Qwen VLo also supports diverse input and output formats, offering increased flexibility for users and making it ideal for creating posters, illustrations, web banners, and social media covers. Alibaba owns the South China Morning Post. The new model adds to the intense competition in China's AI landscape, as rivals such as ByteDance and SenseTime strive to introduce their own multimodal models designed to interpret various types of input data, including text, video, and audio. In contrast, traditional AI models only handle one type of input. 10:41 How Hangzhou's 'Six Little Dragons' built a new Chinese tech hub How Hangzhou's 'Six Little Dragons' built a new Chinese tech hub Alibaba has been doubling down on AI and cloud computing, as it moves to streamline its sprawling operations. In February, the company pledged to invest more than 380 billion yuan (US$52 billion) in AI infrastructure over the next three years.
