What happened
Chinese start-up DeepSeek launched a new large language model (LLM) called DeepSeek V3. This model reportedly outperforms AI products from Meta and OpenAI. The model was trained over two months with a cost of US$5.58 million, using fewer computing resources than competitors.
Why it matters
- DeepSeek's model demonstrates competitive AI capabilities at a lower training cost.
- The model's efficiency could influence the global AI industry by setting new standards for resource use.
- It highlights China's growing presence and innovation in advanced AI technologies.
Key facts
- 1. DeepSeek V3 was trained for two months.
- 2. Training cost was US$5.58 million.
- 3. The model uses significantly fewer computing resources compared to rivals.
