Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.linkup.so/llms.txt

Use this file to discover all available pages before exploring further.

  • Perplexity Sonar is a consumer chatbot wrapped in an API. Linkup is a search API built for production AI.
  • On Verified SimpleQA, Linkup’s Search endpoint ranks first among sub-second web search APIs with a 92% F-score, ahead of Perplexity Sonar Pro, Exa, and Tavily. On SealQA-0, our Research endpoint ranks first across the board with 61% accuracy. The evaluation harness is open source.
  • Linkup serves AI agents and pipelines, not end users. The API returns raw context, sourced answers, or structured JSON, depending on what your code needs.
  • Linkup partners with premium data providers to keep snippet quality and freshness above what general crawlers return.
  • "fast" (beta): our fastest mode with near-instant results. "fast" is a single-pass search directly passed to our index. No query interpretation, reformulation, or evaluation, which makes it ideal for keyword-based search and conversational AI use cases where low latency is critical. Try queries such as “weather in Paris today”. "fast" queries cost 5per1,000querieswithoutputType:"searchResults"(5 per 1,000 queries with `outputType: "searchResults"` (6 per 1,000 with "sourcedAnswer" or "structured").
  • "standard": fast and affordable agentic web search. "standard" is great for quickly finding grounding data from multiple sources or to run AI analysis on a scraped URL. Try queries such as “What is Microsoft’s Q3 2024 revenue?”. "standard" queries cost 5per1,000querieswithoutputType:"searchResults"(5 per 1,000 queries with `outputType: "searchResults"` (6 per 1,000 with "sourcedAnswer" or "structured").
  • "deep": comprehensive, in-depth web search. "deep" handles complex queries that require digging, multi-step information gathering, and reflection. Try queries such as “First, find Linkup website domain. Then scrape the homepage, product page, and about us page. Then search for recent news. Consolidate the output in a comprehensive company profile”. "deep" queries cost 50per1,000querieswithoutputType:"searchResults"(50 per 1,000 queries with `outputType: "searchResults"` (55 per 1,000 with "sourcedAnswer" or "structured").
We post key changes to the API on our Changelog. We also discuss them on our Discord and on X.
You can send us an email at support@linkup.so or ask us directly on Discord.
Linkup runs a proprietary retrieval model trained for high-precision search, combined with selected third-party LLMs for query interpretation and synthesis on "standard" and "deep". The model mix is tuned for accuracy and latency on the SimpleQA and SealQA benchmarks; we change the underlying components when a measurable improvement is available.
If you get this type of error, make sure you have installed the correct package:pip install linkup-sdkYou can reach out to support@linkup.so if you still experience issues.
Need help? Email support@linkup.so, ping us on Discord, or talk to us.