When we think about Chinese AI models, we usually picture benchmark leaderboards and debates about open weights. This week, SenseTime showed something different: its multimodal SenseNova (日日新) model in real, large-scale use at the Olympic Games in Paris. This is not a lab test – it is a deployment seen by millions of viewers around the world.
What exactly is Chinese AI doing at the Olympics?
SenseTime enters the Games through three channels at once. First, broadcasting: InnoMotion, developed with Shanghai Media Tech (SMT), is a 3D motion capture system that works without any sensors or cameras on the athletes. It gives commentators and viewers real-time AR layers – the flight path of an arrow in archery, the spin and direction of a table tennis ball, or the precise bounce point. For the average fan, this is the difference between watching sport and understanding it.
Second, sports analytics: SenseNova powers a movement analysis platform that recognizes athlete positions and object trajectories in real time. The Chinese national basketball team uses it for match analysis and strategy planning, from training all the way through the tournament. Third, fan experiences: generative Olympic-style photo templates and personalized event calendars run on SenseNova for fans worldwide.
Why this matters more than another benchmark record
The Olympics are the biggest global stage imaginable. The Milan Winter Games already announced the first official Olympic large language model in history, built on Alibaba’s Qwen, back in February 2026 – and now Paris is showing SenseNova in action. Chinese AI has stopped proving its value on tests and started proving it in products watched by hundreds of millions of people.
This is a business signal, not just a technical one. Chinese model makers have hit the wall that cloud providers hit before them: a good model alone is not enough, you have to show an application someone will pay for. Sport, media and entertainment are a natural proving ground – they demand multimodality (image, video, text), low latency and live operation. Exactly what SenseNova is doing in Paris.
What it means for the AI market
First, real-time deployments are becoming the new marketing currency. Leaderboard comparisons no longer suffice – clients want to see a model in live action. Second, the Olympics are also a reliability test: a failure during a live broadcast is public embarrassment, so SenseTime had to deliver something that works flawlessly for the entire tournament. Third, this is another step in the global expansion of the Chinese ecosystem – first open weights, then presence at the world’s biggest events.
For anyone choosing models for their own projects, this is a practical hint: multimodal Chinese models are no longer just a curiosity from papers. SenseNova, Qwen or GLM have documented deployments in demanding production environments. It is worth testing their APIs next to Western counterparts before budget decisions are made – especially where live image and video processing matters.
This content was generated using artificial intelligence and reviewed by a human editor.
Sources:
• NetEase Tech – reporting series on SenseNova at the Paris Games (tech.163.com)
• The Paper (澎湃新闻) – SenseTime and SMT: InnoMotion, sports analytics and fan experiences (thepaper.cn)
• AIbase – first official Olympic LLM (Qwen/Alibaba) at the Milan Winter Games (aibase.com)
• SenseTime – official SenseNova platform (sensenova.cn)
• Hermes Agent China AI Weekly Brief, August 2, 2026
This article is also available in Polish (PL).
Leave a Reply
You must be logged in to post a comment.