Published • loading... • Updated
ByteDance targets mega AI model nearing Anthropic’s Mythos, FT reports
The project is in early pre-training and could take three to six months before fine-tuning, as ByteDance seeks to close the gap with frontier labs.
- ByteDance, the owner of TikTok, is training an AI model with around 10 trillion parameters to rival Anthropic's Mythos, according to the Financial Times.
- ByteDance aims to match Western frontier systems as a symbolic milestone, leveraging its TikTok and Doubao platforms to secure a unique distribution channel and training data.
- Scaling to 10 trillion parameters tests how far Chinese firms can push with limited Nvidia chip access, while Moonshot's Kimi K3, a leading Chinese model, has about 2.8 trillion parameters.
- Founder Zhang Yiming instructed staff to build genuine capabilities rather than copying existing models; the project remains in an early pre-training phase expected to take three to six months.
- China has considered restricting overseas access to its best models as these systems become national assets, demonstrating ByteDance's determination to compete at the top.
Insights by Ground AI
Podcasts & Opinions
17 Articles
17 Articles
Jornal ExpressoReposted by 
Observador
ObservadorLeaded by founder Zhang Yiming, ByteDance has strengthened over the past three years investments in data centers, semiconductors and the recruitment of researchers for the AI development team, baptized by Seed and composed of about 2,000 professionals worldwide
·Paço de Arcos, Portugal
Read Full Article
(Seoul = Yonhap News) Reporter Im Hwa-seop = ByteDance, the Chinese tech company and parent of TikTok, [is] a massive ... with potential parameters reaching 10 trillion...
·Seoul, Korea (the Republic of)
Read Full Article
ByteDance targets mega AI model nearing Anthropic’s Mythos, FT reports
Aug 7 (Reuters) - ByteDance is training an AI model that could approach the size of Anthropic's most cutting-edge Mythos system, the Financial Times reported on Friday, citing people with knowledge of the matter. Reuters could not immedia...
Coverage Details
Total News Sources17
Leaning Left1Leaning Right5Center3Last UpdatedBias Distribution56% Right
Bias Distribution
- 56% of the sources lean Right
56% Right
11%
C 33%
R 56%
Factuality
To view factuality data please Upgrade to Premium
