DeepSeek Launches Search and Fetch APIs and Releases Harness v0.2.1-rc1
DeepSeek has officially launched its proprietary Search and Fetch APIs to provide real-time web content for AI agents and RAG applications, alongside releasing the DeepSeek Harness v0.2.1-rc1 update. These new developer tools enable structured web data extraction and real-time browsing capabilities at competitive pricing, enhancing the ecosystem for building autonomous AI agents and retrieval-augmented applications. The Search and Fetch APIs are billed at 5 CNY per thousand successful requests and feature rate limits of 5 queries per second per account. Additionally, a new V4.1 Flash-1017 model snapshot briefly appeared in the API list before being removed.
## BACKGROUND
Retrieval-Augmented Generation (RAG) is a technique that improves the accuracy and reliability of generative AI models by fetching facts from external sources before generating responses. AI agents are autonomous software programs driven by large language models that can make decisions, use external tools, and execute multi-step tasks independently.