Chinese AI company DeepSeek has announced the expansion of its services into the Search API market by making its internal search engine available to external developers through two new APIs: Search and Fetch. This move follows a series of successes the company has achieved in providing low-cost AI models to challenge major companies in the field.
The Search API allows developers to retrieve ranked web search results, including titles, URLs, and summaries, while the Fetch API retrieves web page content in Markdown format. Both services are priced at $0.75 per 1,000 successful requests and operate completely independently of the company's AI models, allowing search capabilities to be integrated into applications running on competing models.
Why does this news matter?
Large Language Models (LLMs) and AI Agents require dynamic and constantly updated information that training data alone cannot provide, such as modern software documentation or newly discovered security vulnerabilities. Previously, using the DeepSeek search engine required an additional call to the AI model, which increased latency and cost. The new APIs eliminate this need and give developers direct access to search results at a low cost.
Infrastructure and Market Evolution
Until recently, DeepSeek relied on third-party providers such as Bocha and Microsoft Bing. However, the company has been working on developing its own search infrastructure since the beginning of the year. Its current engine, called DeepSeek Search, manages hundreds of millions of requests daily and indexes hundreds of billions of web pages, with mechanisms to filter out spam and low-quality AI content.
This launch comes at a time when the market is witnessing increasing competition, with other companies like Perplexity, Brave, and Exa investing in proprietary search infrastructure, especially as giants like Google and Microsoft have withdrawn from offering traditional search APIs openly to new developers.