According to a report by the Financial Times, TikTok owner ByteDance is working on a massive artificial intelligence system designed to rival the most advanced frontier models built in the U.S.
The Beijing-based tech giant is reportedly in the early training phases of a super-sized model that could approach 10 trillion parameters. If successful, the new platform would directly challenge Anthropic’s cutting-edge Mythos model. The sheer scale of this project highlights how aggressively ByteDance is pursuing AI capabilities.
Developing a model with nearly 10 trillion parameters requires immense computing infrastructure, massive energy supplies, and advanced engineering techniques to coordinate thousands of graphics processors working in parallel.
ByteDance has already made significant strides in the artificial intelligence sector over the past year. The company’s Doubao family of models has gained massive popularity across Asia, powering everything from content creation tools to conversational assistants. However, building a system that approaches the raw computational power of Anthropic’s Mythos represents a major step forward for the Chinese tech firm.
The reference point for this effort, Anthropic’s Mythos system, sent shockwaves through the technology industry when details of its capabilities first emerged earlier this year. Frontier systems of that caliber demonstrate unprecedented proficiency in logic, autonomous coding, and complex software vulnerability analysis.
Achieving training runs on this scale presents a substantial technical challenge for Chinese developers. Strict export controls imposed by Washington have restricted direct shipments of high-end graphics processing units to China. To work around those limits, companies have had to rely on alternative hardware setups, distributed computing architectures, and software optimizations to squeeze every drop of performance out of available chips.
As training continues, industry analysts will be watching closely to see if ByteDance can successfully complete the resource-intensive process without running into hardware bottlenecks.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
