> ## Documentation Index > Fetch the complete documentation index at: https://radarkit.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # Get web search > How often the models ran a web search before answering, and what they searched for. The share of answers that searched, then one entry per prompt with the distinct search queries, how many times each was seen, and the average number of queries per search. Use it: pass `prompt_id` for one prompt (what the Web Search card on its page shows), or leave it out for every active prompt, most queries first (the Query Fanout table). Cost: 2 API credits. Limits: Up to 31 days in one call. `limit` caps the prompts returned (default 50, up to 100); `meta.prompts_total` says how many there were. Needs: Read. ## OpenAPI ````yaml /openapi.yaml get /projects/{project}/web-search openapi: 3.1.0 info: title: Radarkit API version: 1.0.0 summary: Your AI visibility data, outside the dashboard. description: > Radarkit tracks how AI assistants answer the questions your customers ask, and what each answer says about your brand. This API gives you that data as JSON. Every request needs an API key in the `Authorization` header. Every response has the same shape: `data`, `meta`, `credits` and `request_id`. Most calls cost 1 API credit. Heavier calls cost 2. Exports spend Export rows instead, one per row. contact: name: Radarkit support email: mail@radarkit.ai url: https://radarkit.ai servers: - url: https://api.radarkit.ai/v1 description: Production security: - bearerAuth: [] tags: - name: Account description: Your key, your plan and both balances. Free. - name: Projects description: One project is one brand or website. Every other call needs a project id. - name: Topics description: Topics group prompts. Every prompt belongs to one topic. - name: Prompts description: The questions Radarkit asks every AI model on a schedule. - name: Visibility description: The daily visibility score per model, precomputed. - name: Responses description: One row per AI answer, with everything Radarkit found in it. - name: Sources description: The web pages AI answers cite or consult. - name: Competitors description: Who the answers recommend and how often. - name: Rankings description: Recommended domains with their position in the answer. - name: Query fanout description: The web searches a model ran before answering, one row per answer. - name: Web search description: How often the models searched and what they searched for, summarised. - name: Sentiment description: How the answers talk about your brand. - name: Exports description: > Thousands of rows as one gzip file. Four calls in order: estimate, create, poll, download. Estimate is free and returns the exact row count. Create charges that many Export rows and queues the job. Poll the export until its status is `ready`. Download follows a redirect to a signed link that works for 30 minutes. Files are kept for 7 days. - name: Content description: Have the Content agent write an article for a prompt, then poll the run. paths: /projects/{project}/web-search: get: tags: - Web search summary: Get web search description: > How often the models ran a web search before answering, and what they searched for. The share of answers that searched, then one entry per prompt with the distinct search queries, how many times each was seen, and the average number of queries per search. Use it: pass `prompt_id` for one prompt (what the Web Search card on its page shows), or leave it out for every active prompt, most queries first (the Query Fanout table). Cost: 2 API credits. Limits: Up to 31 days in one call. `limit` caps the prompts returned (default 50, up to 100); `meta.prompts_total` says how many there were. Needs: Read. operationId: getWebSearch parameters: - $ref: '#/components/parameters/project' - $ref: '#/components/parameters/from' - $ref: '#/components/parameters/to' - $ref: '#/components/parameters/model' - $ref: '#/components/parameters/prompt_id' - $ref: '#/components/parameters/limit' responses: '200': description: The search rate and the queries per prompt. content: application/json: schema: type: object properties: data: $ref: '#/components/schemas/WebSearch' meta: type: object properties: project: type: string from: type: string format: date to: type: string format: date models: type: array items: type: string prompt_id: type: - integer - 'null' limit: type: integer count: type: integer prompts_total: type: integer credits: $ref: '#/components/schemas/Credits' request_id: $ref: '#/components/schemas/RequestId' example: data: answers: 18 searched: 16 search_rate: 88.9 prompts: - prompt: id: 2048 text: best crm for a startup answers: 18 searched: 16 search_rate: 88.9 query_count: 10 avg_queries_per_search: 1.8 queries: - query: best crm startups 2026 count: 3 - query: hubspot free crm limits count: 1 meta: project: 64f1a2b3c4d5e6f708192a3b from: '2026-08-01' to: '2026-08-31' models: - chatgpt - gemini - perplexity prompt_id: 2048 limit: 50 count: 1 prompts_total: 1 credits: charged: 2 remaining: 2934 request_id: req_01j9x4v9l0m1n2o3p4q5r6s7u5 '400': $ref: '#/components/responses/BadWindow' '401': $ref: '#/components/responses/Unauthenticated' '402': $ref: '#/components/responses/PaymentRequired' '403': $ref: '#/components/responses/Forbidden' '404': $ref: '#/components/responses/ProjectNotFound' '422': $ref: '#/components/responses/ValidationFailed' '429': $ref: '#/components/responses/RateLimited' '504': $ref: '#/components/responses/QueryTimeout' components: parameters: project: name: project in: path required: true description: The project id. 24 characters. From List projects. schema: type: string pattern: ^[0-9a-f]{24}$ example: 64f1a2b3c4d5e6f708192a3b from: name: from in: query description: >- First day, `YYYY-MM-DD`, in the project's timezone. Default: the window your dashboard shows, 7 or 30 days back. schema: type: string format: date example: '2026-08-01' to: name: to in: query description: 'Last day, `YYYY-MM-DD`, included. Default: today.' schema: type: string format: date example: '2026-08-31' model: name: model in: query description: >- One model name, or several separated by commas: `chatgpt`, `gemini`, `perplexity`, `copilot`, `aimode`, `aioverview`. Default: every model the project tracks. Names the plan does not include are ignored. schema: type: string maxLength: 120 example: chatgpt,gemini prompt_id: name: prompt_id in: query description: Only rows for this prompt. schema: type: integer minimum: 1 limit: name: limit in: query description: Rows per page, 1 to 100. Default 50. schema: type: integer minimum: 1 maximum: 100 default: 50 schemas: WebSearch: type: object properties: answers: type: integer description: Answers in the window. searched: type: integer description: Answers that ran a web search. search_rate: type: number description: searched as a share of answers, 0 to 100. prompts: type: array items: type: object properties: prompt: type: object properties: id: type: integer text: type: - string - 'null' answers: type: integer searched: type: integer search_rate: type: number query_count: type: integer description: Distinct search queries. avg_queries_per_search: type: number description: Average queries per answer that searched. queries: type: array items: type: object properties: query: type: string count: type: integer description: Answers that ran this query. Credits: type: object description: >- What this call cost and what is left. Also in the `X-RadarKit-Credits-Charged` and `X-RadarKit-Credits-Remaining` headers. properties: charged: type: integer description: API credits spent on this call. remaining: type: - integer - 'null' description: API credits left on the billed account. Null on free calls. RequestId: type: string description: >- A unique id for this call. Also in the `X-Request-Id` header. Quote it when you contact support. example: req_01j9x4m2p7q8r9s0t1u2v3w4x5 Error: type: object description: Every failed call has this shape and no `data`. properties: error: type: object properties: code: type: string description: Stable. Build on this, not on the message. message: type: string description: For people. May change. details: description: >- More when there is more to say. Which field failed, how long to wait, which id already exists. oneOf: - type: object - type: array items: type: object request_id: $ref: '#/components/schemas/RequestId' responses: BadWindow: description: >- `DATE_RANGE_TOO_LARGE`: more than 31 days, use an export. `INVALID_CURSOR`: the cursor does not match these filters, start from the first page. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: DATE_RANGE_TOO_LARGE message: The date range can span at most 31 days. Use exports for more. details: max_days: 31 request_id: req_01j9x4s6b7c8d9e0f1g2h3i4j6 Unauthenticated: description: >- `UNAUTHENTICATED`: no key, a wrong key, or a revoked or expired key. `KEY_IN_QUERY` (400) if the key was in the URL. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: UNAUTHENTICATED message: >- Missing or invalid API key. Send it as: Authorization: Bearer rk_live_YOUR_KEY request_id: req_01j9x4q7z8y9x0w1v2u3t4s5r6 PaymentRequired: description: >- `INSUFFICIENT_CREDITS`: no API credits left, nothing ran. `PLAN_INACTIVE`: the key owner's plan is not active. `OWNER_PLAN_INACTIVE`: the project owner's plan is not active. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: INSUFFICIENT_CREDITS message: Not enough API credits. request_id: req_01j9x4q7z8y9x0w1v2u3t4s5r7 Forbidden: description: >- `SCOPE_MISSING`: the key lacks the permission this call needs. `PROJECT_FORBIDDEN`: the key cannot reach this project. `API_FROZEN`: API access for the account is paused. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: PROJECT_FORBIDDEN message: This key cannot access that project. request_id: req_01j9x4q7z8y9x0w1v2u3t4s5r9 ProjectNotFound: description: '`PROJECT_NOT_FOUND`: no project with that id.' content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: PROJECT_NOT_FOUND message: No project with that id. request_id: req_01j9x4q7z8y9x0w1v2u3t4s5s0 ValidationFailed: description: >- `VALIDATION_FAILED`: a parameter is missing or wrong. `details` lists each field. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: VALIDATION_FAILED message: One or more parameters are invalid. details: - path: from message: The from field must match the format Y-m-d. request_id: req_01j9x4s6b7c8d9e0f1g2h3i4j5 RateLimited: description: >- `RATE_LIMITED`: too many calls. Wait the seconds in `Retry-After` and send the same request again. Nothing was charged. headers: Retry-After: description: Seconds to wait. schema: type: integer content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: RATE_LIMITED message: Too many requests. Slow down and retry. details: retry_after: 21 request_id: req_01j9x4r2s3t4u5v6w7x8y9z0a1 QueryTimeout: description: >- `QUERY_TIMEOUT`: the query took too long. Nothing was charged. Narrow the range or create an export. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: code: QUERY_TIMEOUT message: The query took too long. Narrow the date range or use exports. request_id: req_01j9x4r2s3t4u5v6w7x8y9z0a2 securitySchemes: bearerAuth: type: http scheme: bearer description: >- Your API key. Create one in Settings, API keys. Send it as `Authorization: Bearer rk_live_YOUR_KEY`. ````