Toolofficial publisher

web_search (OpenAI)

Tool : A precise action the AI can call · requires openai-api

Web search hosted within the Responses API.

One parameter and it's enabled; same risk profile as all web tools.

Docs

Overall score

7.3 /10

Security: one caveat, see the card

Utility

9.0

Reliability

6.0

Security

One caveat

Ease

1 click · Comfortable

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Developers building agents or conversational applications often need access to up-to-date information without managing their own search infrastructure. OpenAI's web_search tool addresses this by integrating web search directly into the Responses API, with no third-party service to configure.

In practice, it only takes one parameter in the API call: tools: [{"type": "web_search"}]. The model then decides on its own when to query the web, retrieves relevant results, and incorporates them into its response. The tool is compatible with multiple languages through official SDKs (Python, TypeScript, Go, Java, Ruby, C#) as well as direct HTTP calls.

This tool is aimed at developers already using OpenAI's Responses API who want to enrich their applications with real-time data, for example for news assistants, monitoring tools, or autonomous agents. It is available on recent models such as gpt-6-astra.

It is not suited for those looking for a no-code solution or who do not have access to a paid API account. Each call is billed, which can represent a significant cost at scale.

What it offers

·Activation with a single parameter in the Responses API, with no additional configuration. (paid)
·Compatibility with official SDKs for Python, TypeScript, Go, Java, Ruby, and C#, as well as direct HTTP calls. (paid)
·Native integration into agents built with the Agents API and Agents SDK. (paid)
·The model automatically decides when to trigger a web search based on the request context. (paid)
·Available on recent models such as gpt-6-astra. (paid)
·Can be combined with other tools such as file search, function calling, or remote MCP servers. (paid)

Our tips

Combine web_search with other tools in the same API call to build agents capable of both searching the web and querying your files or functions.
Use the OpenAI CLI with the --transform option to extract the response text directly without additional processing.
Choose the right model based on your budget: the rate is $10 per 1,000 calls on standard models, and $25 on non-reasoning models.
Configure the tool at the agent level in the Agents SDK if you are building a multi-agent system, rather than passing it on each individual request.

Watch out

The tool exposes the network and performs external reads, which implies a risk profile similar to any web tool: retrieved data is not guaranteed to be accurate or current.
Per-call billing can become expensive in high-volume scenarios, especially on non-reasoning models billed at $25 per 1,000 calls.

In practice

Price: Paid · from 9 €
Commercial use: Commercial use: yes
Available on: api, sdk-python, sdk-ts
Host: openai-api
FR interface: No

Security and data

Rating: Security: one caveat, see the card
Capabilities: readswritesexecutesnetwork
Your data: Data with the vendor
Maintenance: Active

Links and repository checked on 25 October 2026 · added on 1 March 2025

How to install web_search (OpenAI)

1. Create or log in to your account at platform.openai.com and retrieve your API key. 2. Install the official SDK for your language (e.g. pip install openai for Python). 3. Set the OPENAI_API_KEY environment variable with your key, without storing it in plain text in your code. 4. In your call to client.responses.create(), add tools=[{"type": "web_search"}] and specify a compatible model such as gpt-6-astra. 5. Retrieve the response via response.output_text and verify cited sources before using them in production.

pip install openai

Official documentation ↗

In the same category

Whole category Feed (memory, context) →