Yandex Cloud
Search
Contact UsTry it for free
  • Customer Stories
  • Documentation
  • Blog
  • All Services
  • System Status
  • Marketplace
    • Featured
    • Infrastructure & Network
    • Data Platform
    • AI for business
    • Security
    • DevOps tools
    • Serverless
    • Monitoring & Resources
  • All Solutions
    • By industry
    • By use case
    • Economics and Pricing
    • Security
    • Technical Support
    • Start testing with double trial credits
    • Cloud credits to scale your IT product
    • Gateway to Russia
    • Cloud for Startups
    • Center for Technologies and Society
    • Yandex Cloud Partner program
    • Price calculator
    • Pricing plans
  • Customer Stories
  • Documentation
  • Blog
© 2026 Direct Cursus Technology L.L.C.
Yandex AI Studio
  • Getting started with Model Gallery
    • About Yandex AI Studio
      • AI Search overview
      • Vector Store search indexes
      • File search tool
      • Web search tool
    • Yandex Workflows
    • Quotas and limits
    • Terms and definitions
  • Switching from the AI Assistant API to Responses API
  • Compatibility with OpenAI
  • Access management
  • Pricing policy
  • Audit Trails events
  • Public materials
  • Release notes

In this article:

  • Possible use cases for the tool
  • Use cases
  1. Concepts
  2. AI Search
  3. Web search tool

Web Search

Written by
Yandex Cloud
Updated at December 23, 2025
  • Possible use cases for the tool
  • Use cases

The web search tools allows the model to draw information from open sources for more accurate and relevant date responses. It can search through the entire internet or within a limited list of websites. You can enable the tool in the Responses API and Realtime API. For the prices, see Yandex AI Studio pricing policy.

With the tool on, the model itself decides whether or not it needs additional information to form a response. To allow the model to search through the internet when generating its responses, specify web_search in your request parameters. When connecting the tool via Responses API, the following optional parameters are available:

  • allowed_domains: Array containing up to five domains to search in. If no domains are listed, the model will search through the entire internet.
  • user_location: Search region limit. Give the region code in the region field if you need to.
  • search_context_size: Size of context yielded by the search. Controls the completeness and detail of the response and token consumption when using the tool. Possible values: low, medium, high. The default value is medium.
Python
response = client.responses.create(
    model=f"gpt://{YANDEX_CLOUD_FOLDER}/{YANDEX_CLOUD_MODEL}",
    input="Create a brief overview of the latest LLM news in 2025, only facts, no specualtion.",
    tools=[
        {
            "type": "web_search",
            "filters": {
                "allowed_domains": [
                    "habr.ru"
                ],
                "user_location": {
                    "region": "213",
                },
            },
            "search_context_size": "medium", # Possible options: low | medium | high
        }
    ],
    temperature=0.3,
    max_output_tokens=1000
)

The response, enriched by the internet search, contains the annotations object. Use it to track and show the data sources to the user. The url field of the annotations object contains the address of the website the information was found on. All other fields of the object are populated with default values for compatibility with OpenAI:

"annotations": [
  {
    "end_index": 0,
    "start_index": 0,
    "title": "",
    "type": "url_citation",
    "url": "www.rbc.ru"
  },
  {
    "end_index": 0,
    "start_index": 0,
    "title": "",
    "type": "url_citation",
    "url": "ria.ru"
  }
]

Possible use cases for the toolPossible use cases for the tool

The web search tool can prove useful in various scenarios:

  • Updating reference information: Creating a summary of app store rule updates for this week.
  • Competitive analysis: Gathering the competitors' innovations and roadmaps based on the information published on their websites.
  • Fact-checking press releases: Verifying quotes and figures prior to publication.
  • Trend radar: Compiling a quick news overview on a topic for a daily or weekly digest.
  • Meeting preparation: Putting together a description of the client company ahead of the conference call.

Use casesUse cases

Creating a text agent with web search

Was the article helpful?

Previous
File search tool
Next
Overview
© 2026 Direct Cursus Technology L.L.C.