DeepSeek Pauses Account Top-Ups Amid Rising Popularity of AI Models Competing with ChatGPT
DeepSeek, a Chinese start-up, has made a bold move by temporarily halting developers from topping up their accounts to access the company's artificial intelligence (AI) models. This decision comes as a clear indication of the overwhelming popularity of DeepSeek's products in the tech industry.
The company recently disabled the top-up button on its platform website, citing "server resource constraints" as the reason for this temporary measure. However, they assured users that existing balances can still be utilised during this period.
DeepSeek's Platform offers application programming interfaces (APIs) that act as a gateway for developers and apps to interact with the start-up's AI models. The Hangzhou-based company gained global recognition after the successful launch of its V3 large language model (LLM) and R1 reasoning model. These models were praised for their capabilities, which were said to be on par with products from ChatGPT creator OpenAI, but at a significantly lower cost.
LLMs are AI models trained to understand human language and perform tasks like generating text or answering questions. On the other hand, reasoning models are designed to handle logical reasoning and inference tasks.
Following the release of DeepSeek's chatbot on January 10, the app experienced a rapid increase in daily active users (DAUs), surpassing 22 million by the end of the month. This placed DeepSeek just behind ChatGPT, which had 53 million DAUs at the time, according to market tracker Aicpb.com.
DeepSeek's chatbot is free to use, while the developer platform offers a time-limited discount rate of $0.28 per 1 million output tokens for the V3 model with a context length of 64,000. This is a significant reduction from the original price of $1.1. In comparison, OpenAI charges $10 per 1 million output tokens for the GPT-4o model, offering a longer context length of 128,000.
China's major cloud service providers, including Alibaba Group Holding, Tencent Holdings, and Huawei Technologies, have made the V3 and R1 models available on their platforms for end users. The success of DeepSeek has led investors to reevaluate China's tech companies, which trade at half the valuation of prominent US tech giants.
Despite its growing reputation in the AI community, DeepSeek has remained tight-lipped about its future plans. The company recently turned away journalists and jobseekers who sought information, indicating a sense of mystery surrounding its next steps.
Two new companies with names similar to DeepSeek were registered in Hong Kong, Deepseek Limited and Deepseek (HK) Limited. However, there is no evidence linking these companies to DeepSeek, and the start-up has not provided any comments on this matter.
DeepSeek temporarily halts account top-ups to access AI models due to overwhelming popularity.
DeepSeek's V3 large language model and R1 reasoning model gain recognition for cost-effective capabilities.
DeepSeek's chatbot attracts over 22 million daily active users, trailing behind ChatGPT.
Source: SCMP