Reinforcement learning

- Train against a fixed web index
- Web search for RL training workloads
- Choose on-prem or a dedicated RL API
SPARSEUP is live - Linkup's powerful open-source sparse embedding model. Try it →
Industry-leading accuracy, in your stack via one API call or deployed inside your own cloud.

Take full control of search. Tune the index, pick the sources, set the freshness.
Web search built for AI applications. Get sourced, cited answers with full-text snippets from across the web. Fast, accurate, and grounded in real data from trusted sources.
from linkup import LinkupClient
client = LinkupClient(
api_key="<your-api-key>", # Or set the LINKUP_API_KEY environment variable
)
# Perform a search query
search_response = client.search(
query="What is Microsoft 2024 revenue?",
depth="standard",
output_type="searchResults",
)
print(search_response)Linkup delivers accuracy, speed, and reliability that set a new standard for web search APIs.
Linkup delivers accuracy, speed, and reliability that set a new standard for web search APIs.
Evaluated on benchmark datasets designed to measure factual accuracy for search-augmented systems on web-grounded, fact-seeking questions.
Trusted by Fortune 500s, frontier AI labs, and government agencies
Deployment modes
Build a dedicated, access-controlled index of your proprietary data, so your AI retrieves the right answers without exposing anything beyond your walls.
Run the full indexing layer inside your own infrastructure for maximum control, compliance, and zero data leaving your environment.
Linkup works with the OpenAI SDK wrapper, LangChain, CrewAI, and dozens of other frameworks. Drop it into your existing stack with zero friction.



“Linkup is how Ava accesses the web at scale, from onboarding to sales enablement”
CEO & Founder at Artisan
How Legora deployed a production-grade web search infrastructure globally
