Web Search

Requesty supports native web search across major providers. Pass a single tool definition and Requesty translates it to the right provider format automatically.

Native web search providers: Vertex AI (Google), Azure OpenAI, OpenAI, Anthropic, xAI, Perplexity

Browse models with web search in the Requesty Console. Look for supports_web_search: true in the List Models response.

Need web search on a model without a native search tool? Try Requesty Web Search (Beta).

How It Works

Each provider implements web search differently. Requesty normalizes these differences behind a single tool definition that works across all three inference endpoints:

Endpoint	Format	Tool definition
`/v1/chat/completions`	OpenAI Chat Completions	`"tools": [{ "type": "web_search" }]`
`/v1/responses`	OpenAI Responses API	`"tools": [{ "type": "web_search" }]`
`/v1/messages`	Anthropic Messages	`"tools": [{ "type": "web_search_20250305", "name": "web_search", "max_uses": 5 }]`

We recommend using streaming ("stream": true) for all web search requests. Streaming delivers citations and content in real time, and avoids the extra latency of waiting for the full search to finish.

Usage

Chat Completions
Responses API
Messages API

POST /v1/chat/completions - Add { "type": "web_search" } to the tools array.

curl https://router.requesty.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
  -d '{
    "model": "anthropic/claude-sonnet-5-5",
    "messages": [
      { "role": "user", "content": "What are the latest news in London today?" }
    ],
    "tools": [{ "type": "web_search" }],
    "stream": true
  }'

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai/v1",
)

stream = client.chat.completions.create(
    model="anthropic/claude-sonnet-5-5",
    messages=[
        {"role": "user", "content": "What are the latest news in London today?"}
    ],
    tools=[{"type": "web_search"}],
    stream=True,
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: 'YOUR_REQUESTY_API_KEY',
  baseURL: 'https://router.requesty.ai/v1',
});

const stream = await client.chat.completions.create({
  model: 'anthropic/claude-sonnet-5-5',
  messages: [
    { role: 'user', content: 'What are the latest news in London today?' }
  ],
  tools: [{ type: 'web_search' }],
  stream: true,
});

for await (const chunk of stream) {
  const content = chunk.choices[0]?.delta?.content;
  if (content) process.stdout.write(content);
}

Just change the model to switch providers:

# Vertex / Gemini
model="vertex/gemini-3.1-pro-preview"

# xAI / Grok
model="xai/grok-4.7"

# Perplexity
model="perplexity/sonar-pro"

Response includes web_search metadata on streamed chunks with citation URLs and titles:

{
  "choices": [{
    "delta": {
      "content": "Based on the search results...",
      "web_search": {
        "content": [
          { "url": "https://example.com/news", "title": "London News" }
        ]
      }
    }
  }]
}

POST /v1/responses - Add { "type": "web_search" } to the tools array.

curl https://router.requesty.ai/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
  -d '{
    "model": "openai-responses/gpt-5.6-sol",
    "input": "What are the latest developments in artificial intelligence?",
    "tools": [{ "type": "web_search" }],
    "stream": true
  }'

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai/v1",
)

response = client.responses.create(
    model="openai-responses/gpt-5.6-sol",
    input="What are the latest developments in artificial intelligence?",
    tools=[{"type": "web_search"}],
    stream=True,
)

for event in response:
    if hasattr(event, 'type') and event.type == 'response.completed':
        for item in event.response.output:
            if item.type == 'message':
                for content in item.content:
                    if content.type == 'output_text':
                        print(content.text)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: 'YOUR_REQUESTY_API_KEY',
  baseURL: 'https://router.requesty.ai/v1',
});

const response = await client.responses.create({
  model: 'openai-responses/gpt-5.6-sol',
  input: 'What are the latest developments in artificial intelligence?',
  tools: [{ type: 'web_search' }],
});

for (const item of response.output) {
  if (item.type === 'message') {
    for (const content of item.content) {
      if (content.type === 'output_text') {
        console.log(content.text);
      }
    }
  }
}

Response includes web_search_call items and url_citation annotations on text content:

{
  "output": [
    {
      "type": "web_search_call",
      "id": "ws_abc123",
      "status": "completed"
    },
    {
      "type": "message",
      "role": "assistant",
      "content": [{
        "type": "output_text",
        "text": "Recent AI developments include...",
        "annotations": [{
          "type": "url_citation",
          "url": "https://example.com/ai-news",
          "title": "AI News Today",
          "start_index": 0,
          "end_index": 35
        }]
      }]
    }
  ]
}

POST /v1/messages - Uses Anthropic’s native web_search_20250305 tool type.

curl https://router.requesty.ai/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_REQUESTY_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "anthropic/claude-sonnet-5-5",
    "max_tokens": 4096,
    "messages": [
      { "role": "user", "content": "What are the latest developments in artificial intelligence?" }
    ],
    "tools": [{
      "type": "web_search_20250305",
      "name": "web_search",
      "max_uses": 5
    }],
    "stream": true
  }'

import anthropic

client = anthropic.Anthropic(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai",
)

with client.messages.stream(
    model="anthropic/claude-sonnet-5-5",
    max_tokens=4096,
    messages=[
        {"role": "user", "content": "What are the latest developments in artificial intelligence?"}
    ],
    tools=[{
        "type": "web_search_20250305",
        "name": "web_search",
        "max_uses": 5,
    }],
) as stream:
    response = stream.get_final_message()

for block in response.content:
    if block.type == "text":
        print(block.text)

import Anthropic from '@anthropic-ai/sdk';

const client = new Anthropic({
  apiKey: 'YOUR_REQUESTY_API_KEY',
  baseURL: 'https://router.requesty.ai',
});

const stream = client.messages.stream({
  model: 'anthropic/claude-sonnet-5-5',
  max_tokens: 4096,
  messages: [
    { role: 'user', content: 'What are the latest developments in artificial intelligence?' }
  ],
  tools: [{
    type: 'web_search_20250305',
    name: 'web_search',
    max_uses: 5,
  }],
});

const response = await stream.finalMessage();

for (const block of response.content) {
  if (block.type === 'text') {
    console.log(block.text);
  }
}

Response includes server_tool_use, web_search_tool_result, and text blocks with citations:

{
  "content": [
    {
      "type": "server_tool_use",
      "id": "srvtoolu_abc123",
      "name": "web_search",
      "input": { "query": "latest AI developments" }
    },
    {
      "type": "web_search_tool_result",
      "tool_use_id": "srvtoolu_abc123",
      "content": [{
        "type": "web_search_result",
        "url": "https://example.com/ai-news",
        "title": "AI News Today",
        "encrypted_content": "..."
      }]
    },
    {
      "type": "text",
      "text": "Recent AI developments include...",
      "citations": [{
        "type": "web_search_result_location",
        "url": "https://example.com/ai-news",
        "title": "AI News Today",
        "cited_text": "Recent breakthroughs in..."
      }]
    }
  ]
}

Web search requires models with tool use support. Check the Model Library to confirm web search support for specific models.

Requesty Web Search (Beta)

Beta. requesty:web_search is in beta. Its behavior and response format may change.

requesty:web_search is a server-side tool: Requesty runs the search, not the provider. Use it to add web search to models that support tool calling but have no native web search tool.

How it works

Requesty replaces requesty:web_search with a web_search function that takes a single query string.
When the model calls web_search, Requesty runs the search and sends the results back to the model as a tool result. Each search returns the top 5 results and a short answer.
Requesty keeps calling the model until it replies without calling web_search. You only receive that final reply.

Usage

Add { "type": "requesty:web_search" } to the tools array. It works on all three endpoints:

Endpoint	API format	Tool definition
`/v1/chat/completions`	OpenAI Chat Completions	`"tools": [{ "type": "requesty:web_search" }]`
`/v1/responses`	OpenAI Responses API	`"tools": [{ "type": "requesty:web_search" }]`
`/v1/messages`	Anthropic Messages	`"tools": [{ "type": "requesty:web_search" }]`

Chat Completions example:

curl https://router.requesty.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
  -d '{
    "model": "openai/gpt-6-luna",
    "messages": [
      { "role": "user", "content": "What is the weather in London today? One sentence." }
    ],
    "tools": [{ "type": "requesty:web_search" }]
  }'

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_REQUESTY_API_KEY",
    base_url="https://router.requesty.ai/v1",
)

response = client.chat.completions.create(
    model="openai/gpt-6-luna",
    messages=[
        {"role": "user", "content": "What is the weather in London today? One sentence."}
    ],
    tools=[{"type": "requesty:web_search"}],
)

print(response.choices[0].message.content)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: 'YOUR_REQUESTY_API_KEY',
  baseURL: 'https://router.requesty.ai/v1',
});

const response = await client.chat.completions.create({
  model: 'openai/gpt-6-luna',
  messages: [
    { role: 'user', content: 'What is the weather in London today? One sentence.' }
  ],
  // The OpenAI SDK types don't include server-side tools
  tools: [{ type: 'requesty:web_search' } as any],
});

console.log(response.choices[0].message.content);

Response is a standard chat completion containing the model’s final answer:

{
  "id": "rqsty-cmpl-0e6e69f6-b47e-49ec-ab25-854f02892a13",
  "object": "chat.completion",
  "model": "gpt-6-luna",
  "choices": [{
    "index": 0,
    "finish_reason": "stop",
    "message": {
      "role": "assistant",
      "content": "The weather in London today is overcast with a temperature of 17.3°C (63.1°F) and a west-southwest wind blowing at 24.5 kph (15.2 mph)."
    }
  }],
  "usage": {
    "prompt_tokens": 2331,
    "completion_tokens": 46,
    "total_tokens": 2377,
    "cost": 0.001006
  }
}

The same tool works on /v1/responses and /v1/messages:

curl https://router.requesty.ai/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
  -d '{
    "model": "openai/gpt-6-luna",
    "input": "What is the weather in London today? One sentence.",
    "tools": [{ "type": "requesty:web_search" }]
  }'

curl https://router.requesty.ai/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
  -d '{
    "model": "anthropic/claude-sonnet-5-5",
    "max_tokens": 1024,
    "messages": [
      { "role": "user", "content": "What is the weather in London today? One sentence." }
    ],
    "tools": [{ "type": "requesty:web_search" }]
  }'

You can mix requesty:web_search with your own function tools. If the model calls one of your tools, Requesty returns the tool call to you as usual.

Streaming

Streaming ("stream": true) works. While a search runs, Requesty sends the SSE comment line : REQUESTY EXECUTING SERVER SIDE TOOL to keep the connection alive. SSE clients, including the OpenAI SDKs, ignore comment lines. During the beta, Requesty buffers each model turn, so the final answer arrives once the model has finished generating rather than token by token.

Pricing

Requesty runs searches through Tavily. Each search is billed at Tavily’s pay-as-you-go rate of

0.008 per credit, and a basic search uses 1 credit ([Tavily API credits](https://docs.tavily.com/documentation/api-credits)). That makes each search

0.008. Search charges are deducted from your balance on top of the model’s token cost.

Limitations

The model must support tool calling.
The model can’t call web_search and one of your own tools in the same turn. If it does, the request fails with The provider executed both a client and a server side tool call. Aborting execution as this is not supported.
You can’t define your own function named web_search in the same request. Requesty rejects it with tool name conflicts with server-side tool: web_search.
Search results and citations aren’t included in the response.
Unknown requesty: tool types are rejected, e.g. unsupported server-side tool type: requesty:foo.

Getting Started

LLM Gateway

Model Capabilities

Analytics & Monitoring

Access Control

Organization

MCP Gateway

How It Works

Usage

Requesty Web Search (Beta)

How it works

Usage

Streaming

Pricing

Limitations

​How It Works

​Usage

​Requesty Web Search (Beta)

​How it works

​Usage

​Streaming

​Pricing

​Limitations

How It Works

Usage

Requesty Web Search (Beta)

How it works

Usage

Streaming

Pricing

Limitations