Toolofficial publisher

google_search (Gemini)

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

Grounds Gemini responses on Google Search, with sources.

The best free or near-free search tool is the Google index.

Docs

Overall score

7.4 /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

Language models work from data frozen in time: they know nothing about events that occurred after their training cutoff. The google_search tool connects Gemini to Google Search in real time, allowing it to answer questions about recent events and cite verifiable sources.

In practice, when you enable this tool in an API call, Gemini analyzes your question, decides whether a search would improve the answer, automatically generates one or more queries, processes the results, and returns a response enriched with citation annotations. Each annotation links a specific passage of text to its source URL.

This tool is aimed at developers building applications through the Gemini API who need factual, up-to-date, and traceable answers. It works across all languages supported by Google Search.

However, it is not designed for direct end-user consumption: it integrates only through the API and requires development skills. Users looking for a ready-to-use interface should look at other Google products.

What it offers

·Real-time grounding: Gemini queries Google Search to answer questions about recent events beyond its training cutoff.
·Inline citations: each passage in the response is linked to its source URL via url_citation annotations with start and end indexes.
·Automatic query generation: the model creates and executes search queries on its own without developer intervention.
·Search suggestions: the response includes an HTML snippet ready to embed for displaying search suggestions in the interface.
·Combination with other tools: google_search can be used alongside the URL Context tool and code execution.
·Combination with custom tools: on Gemini 3 models, it is possible to mix google_search and custom functions (function calling) in the same call.
·Per-query billing (paid): with Gemini 3 models, each search query executed is billed separately; with earlier models, billing is per prompt.

Our tips

Check the url_citation annotations in the response to extract precisely which sentences are supported by which source, using the start_index and end_index fields.
If your question is complex, Gemini may run multiple search queries in a single call: with Gemini 3 models, each query is billed separately, so anticipate the cost for broad prompts.
Combine google_search with the URL Context tool to ground the response on both the public web and specific URLs you provide.
On Gemini 3 models, you can mix google_search with your own custom functions in the same API call for more complex use cases.
Empty search queries are not counted for billing purposes: only non-empty, unique queries are taken into account.

Watch out

Billing varies by model generation: per search query for Gemini 3, per prompt for Gemini 2.5 and earlier models. A prompt that triggers multiple searches may cost more than expected with Gemini 3.
Older models use a different tool (google_search_retrieval) rather than google_search: check compatibility before migrating an existing integration.

In practice

Price: Freemium
Commercial use: Commercial use: yes
Available on: api
Host: gemini-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 December 2024

How to install google_search (Gemini)

1. Obtain a Gemini API key at ai.google.dev and store it securely, never hard-coded in your source code. 2. Install the appropriate SDK (for example, the Python package google-genai or the JavaScript package @google/genai). 3. Initialize the client with your API key via the GEMINI_API_KEY environment variable. 4. Add {"type": "google_search"} to the tools array in your interactions.create call. 5. Check the response for steps of type google_search_call and google_search_result to access executed queries and search suggestions. 6. Iterate over url_citation annotations in model_output content blocks to display sources in your interface.

tools=[{"google_search":{}}]

Official documentation ↗

In the same category

Whole category Feed (memory, context) →