TL;DR: DeepSeek has launched its own web search and page-fetching APIs built specifically for AI agents, pricing them at $0.75 per 1,000 successful calls, putting it in direct competition with the search infrastructure Google, Bing, and startups like Exa have been selling to AI companies.
Every AI agent that needs to answer a question about something that happened after its training cutoff has the same problem: it needs to search the live web, and somebody has to build and sell the plumbing that makes that possible. DeepSeek just decided it wants to be one of the companies selling that plumbing, not just a buyer of it.
What’s actually in the release
The new offering covers two distinct capabilities that agent developers typically need together. A search API returns ranked results for a query, the same basic function Google and Bing have sold through enterprise APIs for years. A page-fetch API retrieves and parses the actual content of a specific URL, stripping out navigation clutter and ads so an AI model can work with clean text instead of raw HTML. Together, they’re the two building blocks almost every “AI agent browses the web” product needs, and until now most developers have had to stitch together separate vendors, or pay Google and Bing’s enterprise rates, to get both.
Pricing at $0.75 per 1,000 successful calls is aggressive. It undercuts the cost structure that’s made search API access a real line-item expense for companies building agent products at scale, and it follows a pattern DeepSeek has stuck to since its first model releases: compete on price aggressively enough that the cost conversation becomes the main reason to switch, even before anyone’s fully tested quality.
- Search API: ranked web results for a query, similar function to existing enterprise search APIs
- Fetch API: clean, parsed page content from a specific URL, no scraping infrastructure required on the developer’s end
- Pricing: $0.75 per 1,000 successful calls
- Target customer: developers building AI agents that need live web access, not end consumers
Why this matters beyond DeepSeek’s own ecosystem
This is a direct shot at the part of Google and Microsoft’s business that’s quietly become more valuable as agentic AI has grown: selling the web itself, packaged and priced, to the AI companies that need to query it. Google and Bing have had a comfortable duopoly on commercial search API access for years, mostly because building and maintaining a web-scale index is extraordinarily expensive and because most AI companies didn’t need agent-grade search access until relatively recently.
DeepSeek clearly already has a working search index of some kind, since it needed one to build its own search-aware chat products. Turning that into a paid API is a relatively small additional engineering lift once the index already exists, and a meaningfully large new revenue line if agent developers bite on the lower price. Whether DeepSeek’s index is comprehensive and fresh enough to compete on quality, not just price, with Google’s decades of crawling infrastructure is the open question nobody outside the company can answer yet from a press release and a pricing page.
This arrives the same week Amazon and Walmart are making opposite bets on whether to let AI shopping agents operate on their own platforms at all, a divergence we’ve covered separately today, underscoring just how much infrastructure is being built, and how many strategic lines are being drawn, around what AI agents are and aren’t allowed to do on the open web right now.
The SEO angle nobody’s talking about yet
If DeepSeek’s search index becomes a meaningful source of traffic for AI agents doing research and summarization, site owners who’ve spent years optimizing for Google’s crawler conventions may eventually need to think about a second set of rules, how DeepSeek’s crawler identifies itself, whether it respects robots.txt the way Googlebot does, and whether getting indexed there requires anything different than standard technical SEO. None of that infrastructure or documentation is mature yet, which is normal for a product this new, but it’s worth watching if your traffic increasingly comes from AI assistants rather than a traditional search results page.
We covered DeepSeek’s open-source inference engine, DeepInfer, earlier this month, and this search API launch continues a pattern: DeepSeek keeps shipping developer-facing infrastructure, not just consumer chat products, at a pace that’s forcing competitors to respond on price even when they can’t yet match it on polish.
How the incumbents are likely to respond
Google and Microsoft aren’t going to match a $0.75 per 1,000 calls price point without a fight, but they also aren’t going to panic over it the way a smaller startup might. Both companies can afford to subsidize enterprise search API pricing for a long stretch if they decide market share in the agent-infrastructure layer matters more than near-term margin on this specific product line, the same way both companies have historically used pricing as a weapon in cloud infrastructure wars where the real prize was long-term platform lock-in rather than the specific line item being discounted. Exa and other smaller, agent-focused search API startups are in a tougher spot, since they don’t have a sprawling, profitable core business to cross-subsidize a price war the way the giants do.
The more interesting long-term question is whether price ends up being the deciding factor at all once developers actually compare result quality and freshness across providers in production. A cheap search API that returns stale or incomplete results isn’t actually cheap once a developer factors in the engineering time spent compensating for it. DeepSeek’s pricing gets attention today. Whether its index holds up against sustained, large-scale agent traffic is the test that determines whether this becomes a real competitive threat or just an interesting headline.
The bottom line
DeepSeek just made it meaningfully cheaper to build an AI agent that can actually browse the live web, and that alone is enough to force a pricing conversation at Google and Microsoft regardless of how the quality comparison eventually shakes out.
The Teck Hustlers Editorial Team covers technology, business, finance and gaming news for readers who want the story behind the headline. Every article is researched, written and edited by our in-house team, checked against company announcements, filings and reputable reporting, and updated when new facts emerge. We focus on what each development means for founders, professionals, investors and everyday users. Have a tip or a correction? Reach us through our contact page.
