Chinese AI developer DeepSeek has announced that the training of its R1 model cost only $294,000. This figure is significantly lower than the reported costs for similar AI models developed by U.S. companies, potentially sparking renewed discussions about China's role in the AI development race.
The cost estimate, revealed in a peer-reviewed article published in the academic journal Nature, marks a rare update from the Hangzhou-based company. DeepSeek's earlier release of lower-cost AI systems in January had already caused concern among global investors, who worried that these models could challenge the dominance of established AI leaders like Nvidia. Following this initial release, the company and its founder, Liang Wenfeng, have maintained a relatively low public profile, focusing on product updates.
The Nature article, with Liang Wenfeng listed as a co-author, details that the R1 model, which focuses on reasoning, was trained using 512 Nvidia H800 chips. In contrast, OpenAI CEO Sam Altman stated in 2023 that the training of foundational models cost "much more" than $100 million, although the company has not provided specific figures for its releases. DeepSeek's claim, if accurate, suggests a potentially significant cost advantage in AI development for the Chinese company.
Artículos relacionados de LaRebelión:
- Boosting Cyber Awareness Why Employee Training is Key to Preventing Cyber Attacks
- Huawei Chips Stall DeepSeek R2 AI Model A Tech Setback
- AI2s MolmoAct A 3D Robotics AI Model Challenging Nvidia and Google
- DeepMinds Genie 3 Creating Real-Time Interactive Simulations with a New World Model
Artículo generado mediante LaRebelionBOT
No hay comentarios:
Publicar un comentario