DeepSeek Web Search Temporarily Unavailable Due to Technical Issues
DeepSeek's web search feature has become temporarily unavailable due to technical reasons, restricting the model's responses to its internal knowledge cutoff of May 2025. This outage temporarily prevents users from accessing real-time information through the AI, highlighting the reliance of modern LLMs on external search integration for up-to-date queries. While the web search is offline, DeepSeek is also preparing for the rollout of its V4 model, which will introduce a peak-valley API pricing system to optimize GPU resource allocation during busy periods.
## BACKGROUND
Large language models typically rely on static training data, which limits their knowledge to a specific cutoff date. To overcome this, systems use Retrieval-Augmented Generation (RAG) or web search integration to fetch real-time information from external sources before generating a response. DeepSeek has previously integrated its models with search engines like Baidu and Tencent Cloud to support these live queries.