Read more

We’re excited to announce a new partnership with Linkup, bringing their high‑quality, AI‑powered web search directly into the Friendli Suite. Starting today, Linkup Search API is available as a built-in tool in the Serverless Endpoints playground.

Why It Matters

Large language models can’t know what happened five minutes ago, but production AI agents need fresh, factual, and source-backed data. With Linkup now inside Friendli Serverless Endpoints, developers get real-time, trustworthy search results in a single integration, fully compatible across FriendliAI’s AI models. Linkup ranks #1 on OpenAI’s SimpleQA benchmark, so whether you’re building agents for customer support, news, or market insights, reliable answers are now just a tool call away.

This integration helps teams build more reliable, production-ready AI agents by grounding LLM outputs with up‑to‑date information from trusted sources.

About Linkup

Linkup provides an AI‑powered web search API designed specifically for AI agents and LLMs, enabling access to accurate, structured, and source-cited web content to enhance AI performance and reduce hallucinations.

Linkup stands out as the world’s best search for AI apps, achieving the highest score on OpenAI's SimpleQA factuality benchmark.

Getting Started

In the Playground

All you need to do is open the Serverless Endpoints playground and enable the linkup:search tool.

Figure 1: Enabling Linkup Search API tool in the Playground.

That’s it! You can now prompt your models to trigger the Linkup web search tool calls.

Figure 2: Example usage.

Via API and SDK

For API and SDK access, you’ll need a Linkup API key. To enable Linkup’s web search tool:

1. Go to https://app.linkup.so, and get your Linkup API key.
2. In Friendli Suite, open “Personal settings > Integrations” and add your Linkup API key.

Figure 3: Integration settings.

You can now send requests via API and SDK.

bash

curl --request POST \
--url https://api.friendli.ai/serverless/tools/v1/chat/completions \ --header 'Content-Type: application/json' \ --header 'Authorization: Bearer <FRIENDLI_TOKEN>' \ --data '{
"model": "meta-llama-3.1-8b-instruct",
"messages": [
{
"content": "Find information on the popular movies currently showing in theaters and provide their ratings.",
"role": "user"
}
],
"tools": [
{ "type": "linkup:search" }
]
}'

python

import os

from friendli import SyncFriendli, models

TOKEN = os.environ["FRIENDLI_TOKEN"]
TEAM_ID = os.environ["FRIENDLI_TEAM"]

with SyncFriendli(
token=TOKEN,
x_friendli_team=TEAM_ID,
) as friendli:
res = friendli.serverless.tool_assisted_chat.stream(
model="meta-llama-3.1-8b-instruct",
messages=[
{
"role": "user",
"content": "Find information on the popular movies currently showing in"
" theaters and provide their ratings.",
}
],
tools=[{"type": "linkup:search"}],
)
for chunk in res:
if isinstance(chunk.data, models.StreamedChatData):
print(chunk.data.choices[0].delta.content, end="")

For more information, please refer to our documentation and guide in addition to Linkup’s documentation.

Looking Ahead

At FriendliAI, we're committed to delivering the best AI inference platform for powering real-world AI applications. We welcome partnerships as part of our ongoing effort to provide even greater value and better services to our users.

This partnership represents another step forward in our mission to make AI more accessible, efficient, and impactful. By combining FriendliAI’s cutting-edge AI inference technology with Linkup’s advanced AI web search API, we're enabling smarter, faster, and more reliable AI experiences for developers and businesses.

Stay tuned — we've got more to come soon.