DeepSeek search and fetch – $0.75 per 1000 successful requests
DeepSeek has launched new Search and Fetch APIs, giving AI agents and RAG applications real-time web access at a highly competitive rate of $0.75 per 1,000 successful requests. This move challenges existing, often pricier, web scraping and search API services, sparking discussion among developers building with LLMs on the economics and practicalities of integrating live web data.
The Lowdown
DeepSeek has rolled out its new Search and Fetch APIs, providing AI agents and Retrieval-Augmented Generation (RAG) applications with access to live web content. These APIs are designed to be both powerful and cost-effective, leveraging DeepSeek's in-house search capabilities.
- Search API: This API allows users to query the web, powered by DeepSeek's proprietary search engine, and receive structured search results. It supports various advanced query operators like
site:,lang:,before:, andafter:, enabling highly specific searches. - Fetch API: Complementing the Search API, the Fetch API retrieves the full content of a specified web page, returning it in a clean Markdown format for easy consumption by models.
- Pricing Structure: Both APIs are billed at a flat rate of $0.75 per 1,000 successful requests, meaning only requests that return HTTP 200 with actual content are charged. Failed or empty results are not billed, making it an attractive option for developers.
- Authentication & Rate Limits: Access requires a DeepSeek API key, and each API has an independent QPS limit of 5 requests per second at the account level.
- Future Billing: Server-side tool calls through DeepSeek Harness, Claude Code, or Codex that utilize these APIs are currently free, with per-request billing set to commence on November 1, 2026.
These APIs position DeepSeek as a significant player in providing essential web access tools for the rapidly evolving AI ecosystem, potentially lowering the barrier to entry for developers looking to build more informed and up-to-date AI applications.
The Gossip
Pricing Power Plays
Commenters were quick to compare DeepSeek's $0.75/1000 requests to competitors, noting it significantly undercuts others like Brave ($5), Kagi ($12), Exa ($4), and SERP API ($1.50). Many praised the competitive pricing as a huge advantage, especially for those building RAG applications. Some users shared experiences with building their own, more cost-effective scraping solutions, while others warned that AI service prices are likely to 'skyrocket eventually' as companies seek profitability.
Bot Battles and Scraper Shenanigans
A lively debate erupted over the efficacy and ethics of web scraping services, particularly concerning anti-bot protections like Cloudflare and Anubis. Users questioned how DeepSeek handles these challenges, with some suggesting that successful large-scale scraping often requires sophisticated setups (e.g., patched browsers, residential proxies). Others criticized professional scrapers for escalating the 'arms race' that degrades the internet experience for regular users, highlighting that if there's profit to be made, someone will inevitably find a way around protections.
DeepSeek's Dexterity and Market Dynamics
The discussion extended to DeepSeek's overall model performance, with some claiming its Flash model is 'neck and neck' with SOTA models like Anthropic's Opus 5.5, particularly for coding tasks, while others argued it's 'about half a gen behind' but highly cost-effective. The broader competitive landscape was also explored, with speculation on whether DeepSeek's offerings came at a 'bad time' for Anthropic's IPO and the potential for a 'duopoly' or wider market competition among AI giants like OpenAI, Anthropic, Google, and Meta.