AI Gateway adds DeepSeek Search and Fetch: one key, two tools, and no swap when you change model providers
Synopsis
Vercel's AI Gateway now offers DeepSeek Search and DeepSeek Fetch: the first returns relevant pages with titles, URLs and summaries, and the second retrieves page content as Markdown for a model to read. Both work with any model that supports tool calling under a single AI Gateway API key, with no separate DeepSeek account, and the same tools stay in place when switching model providers. Search and Fetch each cost $0.75 per 1,000 billable requests; failed requests, timeouts and empty results carry no tool charge, and model tokens are billed separately.
Interpretation
AI Gateway adds two tools, DeepSeek Search and DeepSeek Fetch, so a model can find pages and then read their content within one call. The two tools sit on AI Gateway, share one AI Gateway API key, require no separate DeepSeek account, and do not change when the model provider changes. Both tools work with any model that supports tool calling and are called with that same key.
Search returns relevant pages with titles, URLs and summaries, while Fetch retrieves page content as Markdown for the model to read. The two tools are passed into tools as gateway.tools.deepseekSearch and gateway.tools.deepseekFetch, and can be used together or paired with alternative search tools when DeepSeek Search fails. These helpers are available in AI SDK 6.0.321 and later on v6, or 7.0.160 and later on v7, updated with pnpm add ai@latest and configured with AI_GATEWAY_API_KEY.
Perspective
The tools target developers who call models through AI Gateway and need web search and page retrieval, and they apply to models that support tool calling on AI SDK 6.0.321 and later on v6 or 7.0.160 and later on v7.
Search fallback is model-directed, so provider order and retry limits are best effort; failed requests, timeouts and empty results carry no tool charge, and model tokens are billed separately.