DeepSeek users encounter new limits on long chats and daily use
Some users of China’s DeepSeek service have reported new limits on continuing long conversations, daily use and text entered in one prompt. Messages attribute the restrictions to insufficient computing capacity. The company had announced no uniform allowance or restoration timetable when the reports appeared; earlier service interruptions and a surge in third-party platform use provide context.
Artificial Intelligence··Evening
Long conversations stop resuming
Some users of Chinese AI company DeepSeek’s app and website have reported restrictions on continuing long conversations and on daily usage. Messages seen by users attribute the limits to insufficient computing capacity. The reports appeared on November 16 after complaints the previous day. People trying to reopen long conversations inactive for more than 7 days encountered a message saying these sessions could no longer continue.[1], [2]
Text inputs and daily allowances are also restricted
Users also reported a single-input limit of about 40,000 Chinese characters. Those reaching their daily allowance saw a separate notice saying the servers were busy and their usage limit had been reached. When IT Home published its account, DeepSeek had issued no standalone announcement defining the new restrictions or giving a restoration timetable. The reported character limit concerns text entry, rather than a token count defining the model’s technical context capacity.[1]
Earlier interruptions accompany heavy platform demand
The company’s status page listed 8 interruptions since October affecting chat and its V4 Pro and V4.1 Flash developer interfaces, according to IT Home. Separately, OpenRouter’s October 19–25 weekly ranking put DeepSeek V4.1 Flash usage at 33.6 trillion tokens, up 801 per cent from the preceding week. That measure covers the third-party platform’s calls. It does not represent all DeepSeek users or the company’s entire computing infrastructure.[1]