How long does this connect openwebui to a search api tutorial take?

Most developers complete this tutorial in 15 to 30 minutes. You will need a Scavio API key (free tier works) and a working Python or JavaScript environment.

What do I need before starting?

OpenWebUI installed (v0.4+). Scavio API key. Admin access to OpenWebUI. A Scavio API key gives you 50 free credits on signup.

Can I run this tutorial with the free tier?

Yes. The free tier includes 50 credits on signup, which is more than enough to complete this tutorial and prototype a working solution.

What frameworks does this work with?

Scavio has a native LangChain package (langchain-scavio), an MCP server, and a plain REST API that works with any HTTP client. This tutorial uses the raw REST API, but you can adapt to your framework of choice.

Connect OpenWebUI to Search API Instead of SearXNG

You can connect OpenWebUI to the Scavio search API by creating a custom Function tool in OpenWebUI's function editor. This replaces the default SearXNG integration with a reliable API-backed search that covers Google, Reddit, YouTube, and Amazon.

Prerequisites

OpenWebUI installed (v0.4+)
Scavio API key
Admin access to OpenWebUI

Walkthrough

Step 1: Open the Function editor in OpenWebUI

Go to Workspace > Functions > New Function. Select Tool as the function type.

Bash

# Navigate to:
# Settings > Workspace > Functions > + (Add Function)
# Function Type: Tool
# Name: web_search
# Description: Search the web using Scavio

Step 2: Write the function tool code

Paste this Python function into the OpenWebUI function editor. It calls the Scavio API and returns formatted results.

Python

import requests
from pydantic import BaseModel, Field

# Scavio has one endpoint per platform -- there is no dispatcher endpoint and no
# "platform" request param, so the selector lives in this function.
ENDPOINTS = {
    "google": ("/api/v2/google", "query"),
    "reddit": ("/api/v1/reddit/search", "query"),
    "youtube": ("/api/v1/youtube/search", "search"),
    "amazon": ("/api/v1/amazon/search", "query"),
}
RESULT_KEYS = ("organic_results", "results", "products")


class Tools:
    class Valves(BaseModel):
        SCAVIO_API_KEY: str = Field(default="", description="Your Scavio API key")

    def __init__(self):
        self.valves = self.Valves()

    def web_search(
        self,
        query: str,
        platform: str = "google"
    ) -> str:
        """
        Search the web for current information.
        :param query: The search query
        :param platform: Platform to search: google, reddit, youtube, amazon
        :return: Formatted search results
        """
        path, query_key = ENDPOINTS.get(platform, ENDPOINTS["google"])
        try:
            r = requests.post(
                "https://api.scavio.dev" + path,
                json={query_key: query},
                headers={
                    "Authorization": f"Bearer {self.valves.SCAVIO_API_KEY}",
                    "Content-Type": "application/json",
                },
                timeout=15
            )
            r.raise_for_status()
            payload = r.json()
            # Google v2 passes Google's response through as-is; the other
            # endpoints wrap their payload in a "data" object.
            body = payload.get("data", payload)
            results = next((body[k] for k in RESULT_KEYS if body.get(k)), [])
            if not results:
                return "No results found."
            lines = []
            for i, res in enumerate(results[:5], 1):
                title = res.get("title") or res.get("name", "")
                link = res.get("link") or res.get("url", "")
                snippet = res.get("snippet") or res.get("text", "")
                lines.append(f"{i}. {title}\n   {snippet}\n   {link}")
            return "\n\n".join(lines)
        except Exception as e:
            return f"Search error: {str(e)}"

Step 3: Configure the API key in Valve settings

After saving the function, click the Valve icon on the function card and enter your Scavio API key in SCAVIO_API_KEY.

Bash

# In OpenWebUI:
# Functions > web_search > Valve Settings (wrench icon)
# SCAVIO_API_KEY: your-scavio-api-key
# Save

Step 4: Enable the tool in a model chat

Start a new chat, click the tools icon, and enable 'web_search'. Test it with a query that requires current data.

Bash

# Test prompt:
# "What are the latest AI model releases in 2026? Search the web and summarize."

# Expected behavior:
# OpenWebUI calls web_search(query="latest AI model releases 2026")
# Returns top 5 results formatted as numbered list
# Model synthesizes the results into an answer

Python Example

Python

# Standalone test of the function logic before loading into OpenWebUI
import requests

SCAVIO_KEY = "your-scavio-api-key"

# One endpoint per platform -- Scavio has no dispatcher endpoint and no
# "platform" request param.
ENDPOINTS = {
    "google": ("/api/v2/google", "query"),
    "reddit": ("/api/v1/reddit/search", "query"),
    "youtube": ("/api/v1/youtube/search", "search"),
    "amazon": ("/api/v1/amazon/search", "query"),
}
RESULT_KEYS = ("organic_results", "results", "products")

def web_search(query: str, platform: str = "google", num_results: int = 5) -> str:
    path, query_key = ENDPOINTS.get(platform, ENDPOINTS["google"])
    try:
        r = requests.post(
            "https://api.scavio.dev" + path,
            json={query_key: query},
            headers={
                "Authorization": f"Bearer {SCAVIO_KEY}",
                "Content-Type": "application/json",
            },
            timeout=15
        )
        r.raise_for_status()
        payload = r.json()
        # Google v2 is a raw passthrough; the rest wrap their payload in "data".
        body = payload.get("data", payload)
        results = next((body[k] for k in RESULT_KEYS if body.get(k)), [])
        if not results:
            return "No results found."
        lines = []
        for i, res in enumerate(results[:num_results], 1):
            title = res.get("title") or res.get("name", "")
            link = res.get("link") or res.get("url", "")
            snippet = res.get("snippet") or res.get("text", "")
            lines.append(f"{i}. {title}\n   {snippet}\n   {link}")
        return "\n\n".join(lines)
    except Exception as e:
        return f"Search error: {str(e)}"

# Test before deploying to OpenWebUI
if __name__ == "__main__":
    print(web_search("latest AI models 2026"))
    print("\n" + "="*50 + "\n")
    print(web_search("Claude API pricing", platform="reddit"))

JavaScript Example

JavaScript

// Test the API call from Node.js before configuring in OpenWebUI
const SCAVIO_KEY = 'your-scavio-api-key';

// One endpoint per platform -- Scavio has no dispatcher endpoint and no
// "platform" request param.
const ENDPOINTS = {
  google: ['/api/v2/google', 'query'],
  reddit: ['/api/v1/reddit/search', 'query'],
  youtube: ['/api/v1/youtube/search', 'search'],
  amazon: ['/api/v1/amazon/search', 'query'],
};
const RESULT_KEYS = ['organic_results', 'results', 'products'];

async function webSearch(query, platform = 'google', numResults = 5) {
  const [path, queryKey] = ENDPOINTS[platform] ?? ENDPOINTS.google;
  const res = await fetch('https://api.scavio.dev' + path, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      Authorization: `Bearer ${SCAVIO_KEY}`,
    },
    body: JSON.stringify({ [queryKey]: query })
  });
  if (!res.ok) return `Search error: HTTP ${res.status}`;
  const payload = await res.json();
  // Google v2 is a raw passthrough; the rest wrap their payload in "data".
  const body = payload.data ?? payload;
  const results = RESULT_KEYS.map(k => body[k]).find(v => v?.length) ?? [];
  if (!results.length) return 'No results found.';
  return results.slice(0, numResults).map((r, i) =>
    `${i + 1}. ${r.title ?? r.name ?? ''}\n   ${r.snippet ?? r.text ?? ''}\n   ${r.link ?? r.url ?? ''}`
  ).join('\n\n');
}

console.log(await webSearch('latest AI models 2026'));

Expected Output

JSON

1. Anthropic Releases Claude 4 Opus with Extended Thinking
   Anthropic's latest model Claude 4 Opus introduces extended thinking mode and 200K context window...
   https://anthropic.com/news/claude-4-opus

2. OpenAI GPT-5 Benchmark Results 2026
   GPT-5 achieves state-of-the-art results on...
   https://openai.com/research/gpt-5

Prerequisites

OpenWebUI installed (v0.4+)
Scavio API key
Admin access to OpenWebUI

Walkthrough

Step 1: Open the Function editor in OpenWebUI

Go to Workspace > Functions > New Function. Select Tool as the function type.

Bash

# Navigate to:
# Settings > Workspace > Functions > + (Add Function)
# Function Type: Tool
# Name: web_search
# Description: Search the web using Scavio

Step 2: Write the function tool code

Paste this Python function into the OpenWebUI function editor. It calls the Scavio API and returns formatted results.

Python

import requests
from pydantic import BaseModel, Field

# Scavio has one endpoint per platform -- there is no dispatcher endpoint and no
# "platform" request param, so the selector lives in this function.
ENDPOINTS = {
    "google": ("/api/v2/google", "query"),
    "reddit": ("/api/v1/reddit/search", "query"),
    "youtube": ("/api/v1/youtube/search", "search"),
    "amazon": ("/api/v1/amazon/search", "query"),
}
RESULT_KEYS = ("organic_results", "results", "products")


class Tools:
    class Valves(BaseModel):
        SCAVIO_API_KEY: str = Field(default="", description="Your Scavio API key")

    def __init__(self):
        self.valves = self.Valves()

    def web_search(
        self,
        query: str,
        platform: str = "google"
    ) -> str:
        """
        Search the web for current information.
        :param query: The search query
        :param platform: Platform to search: google, reddit, youtube, amazon
        :return: Formatted search results
        """
        path, query_key = ENDPOINTS.get(platform, ENDPOINTS["google"])
        try:
            r = requests.post(
                "https://api.scavio.dev" + path,
                json={query_key: query},
                headers={
                    "Authorization": f"Bearer {self.valves.SCAVIO_API_KEY}",
                    "Content-Type": "application/json",
                },
                timeout=15
            )
            r.raise_for_status()
            payload = r.json()
            # Google v2 passes Google's response through as-is; the other
            # endpoints wrap their payload in a "data" object.
            body = payload.get("data", payload)
            results = next((body[k] for k in RESULT_KEYS if body.get(k)), [])
            if not results:
                return "No results found."
            lines = []
            for i, res in enumerate(results[:5], 1):
                title = res.get("title") or res.get("name", "")
                link = res.get("link") or res.get("url", "")
                snippet = res.get("snippet") or res.get("text", "")
                lines.append(f"{i}. {title}\n   {snippet}\n   {link}")
            return "\n\n".join(lines)
        except Exception as e:
            return f"Search error: {str(e)}"

Step 3: Configure the API key in Valve settings

After saving the function, click the Valve icon on the function card and enter your Scavio API key in SCAVIO_API_KEY.

Bash

# In OpenWebUI:
# Functions > web_search > Valve Settings (wrench icon)
# SCAVIO_API_KEY: your-scavio-api-key
# Save

Step 4: Enable the tool in a model chat

Start a new chat, click the tools icon, and enable 'web_search'. Test it with a query that requires current data.

Bash

# Test prompt:
# "What are the latest AI model releases in 2026? Search the web and summarize."

# Expected behavior:
# OpenWebUI calls web_search(query="latest AI model releases 2026")
# Returns top 5 results formatted as numbered list
# Model synthesizes the results into an answer

Python Example

Python

# Standalone test of the function logic before loading into OpenWebUI
import requests

SCAVIO_KEY = "your-scavio-api-key"

# One endpoint per platform -- Scavio has no dispatcher endpoint and no
# "platform" request param.
ENDPOINTS = {
    "google": ("/api/v2/google", "query"),
    "reddit": ("/api/v1/reddit/search", "query"),
    "youtube": ("/api/v1/youtube/search", "search"),
    "amazon": ("/api/v1/amazon/search", "query"),
}
RESULT_KEYS = ("organic_results", "results", "products")

def web_search(query: str, platform: str = "google", num_results: int = 5) -> str:
    path, query_key = ENDPOINTS.get(platform, ENDPOINTS["google"])
    try:
        r = requests.post(
            "https://api.scavio.dev" + path,
            json={query_key: query},
            headers={
                "Authorization": f"Bearer {SCAVIO_KEY}",
                "Content-Type": "application/json",
            },
            timeout=15
        )
        r.raise_for_status()
        payload = r.json()
        # Google v2 is a raw passthrough; the rest wrap their payload in "data".
        body = payload.get("data", payload)
        results = next((body[k] for k in RESULT_KEYS if body.get(k)), [])
        if not results:
            return "No results found."
        lines = []
        for i, res in enumerate(results[:num_results], 1):
            title = res.get("title") or res.get("name", "")
            link = res.get("link") or res.get("url", "")
            snippet = res.get("snippet") or res.get("text", "")
            lines.append(f"{i}. {title}\n   {snippet}\n   {link}")
        return "\n\n".join(lines)
    except Exception as e:
        return f"Search error: {str(e)}"

# Test before deploying to OpenWebUI
if __name__ == "__main__":
    print(web_search("latest AI models 2026"))
    print("\n" + "="*50 + "\n")
    print(web_search("Claude API pricing", platform="reddit"))

JavaScript Example

JavaScript

// Test the API call from Node.js before configuring in OpenWebUI
const SCAVIO_KEY = 'your-scavio-api-key';

// One endpoint per platform -- Scavio has no dispatcher endpoint and no
// "platform" request param.
const ENDPOINTS = {
  google: ['/api/v2/google', 'query'],
  reddit: ['/api/v1/reddit/search', 'query'],
  youtube: ['/api/v1/youtube/search', 'search'],
  amazon: ['/api/v1/amazon/search', 'query'],
};
const RESULT_KEYS = ['organic_results', 'results', 'products'];

async function webSearch(query, platform = 'google', numResults = 5) {
  const [path, queryKey] = ENDPOINTS[platform] ?? ENDPOINTS.google;
  const res = await fetch('https://api.scavio.dev' + path, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      Authorization: `Bearer ${SCAVIO_KEY}`,
    },
    body: JSON.stringify({ [queryKey]: query })
  });
  if (!res.ok) return `Search error: HTTP ${res.status}`;
  const payload = await res.json();
  // Google v2 is a raw passthrough; the rest wrap their payload in "data".
  const body = payload.data ?? payload;
  const results = RESULT_KEYS.map(k => body[k]).find(v => v?.length) ?? [];
  if (!results.length) return 'No results found.';
  return results.slice(0, numResults).map((r, i) =>
    `${i + 1}. ${r.title ?? r.name ?? ''}\n   ${r.snippet ?? r.text ?? ''}\n   ${r.link ?? r.url ?? ''}`
  ).join('\n\n');
}

console.log(await webSearch('latest AI models 2026'));

Expected Output

JSON

1. Anthropic Releases Claude 4 Opus with Extended Thinking
   Anthropic's latest model Claude 4 Opus introduces extended thinking mode and 200K context window...
   https://anthropic.com/news/claude-4-opus

2. OpenAI GPT-5 Benchmark Results 2026
   GPT-5 achieves state-of-the-art results on...
   https://openai.com/research/gpt-5

How to Connect OpenWebUI to a Search API

Prerequisites

Walkthrough

Step 1: Open the Function editor in OpenWebUI

Step 2: Write the function tool code

Step 3: Configure the API key in Valve settings

Step 4: Enable the tool in a model chat

Python Example

JavaScript Example

Expected Output

Related Tutorials

Frequently Asked Questions

How long does this connect openwebui to a search api tutorial take?

What do I need before starting?

Can I run this tutorial with the free tier?

What frameworks does this work with?

Related Resources

OpenWebUI Search Integration with Search API

OpenWebUI Reliable Search Backend

Best Search APIs for OpenWebUI in May 2026

Best Search API for OpenWebUI Tools in 2026

Add Unified Search to Multi-Agent Systems with Scavio

MCP Search Integration vs Direct API Integration

Start Building

How to Connect OpenWebUI to a Search API

Prerequisites

Walkthrough

Step 1: Open the Function editor in OpenWebUI

Step 2: Write the function tool code

Step 3: Configure the API key in Valve settings

Step 4: Enable the tool in a model chat

Python Example

JavaScript Example

Expected Output

Related Tutorials

Frequently Asked Questions

How long does this connect openwebui to a search api tutorial take?

What do I need before starting?

Can I run this tutorial with the free tier?

What frameworks does this work with?

Related Resources

OpenWebUI Search Integration with Search API

OpenWebUI Reliable Search Backend

Best Search APIs for OpenWebUI in May 2026

Best Search API for OpenWebUI Tools in 2026

Add Unified Search to Multi-Agent Systems with Scavio

MCP Search Integration vs Direct API Integration

Start Building