Pricing and Billing

    Pricing and Billing

    How billing works, where to see per-request cost in USD, and how pricing is grouped.

    Billing

    Manage payments, balance, request logs, and spending from the Billing menu in Desearch Console.

    Supported top-up methods:

    • Card payments: add funds with a credit or debit card.
    • TAO cryptocurrency payments: add funds using TAO where supported.

    You can enable auto-recharge with a linked card so your balance is topped up when it reaches a configured threshold. Console also provides transaction history, request-cost breakdowns, and usage visibility by API key/service.

    Per-request cost in USD

    Billable API responses expose the exact request cost in USD. Use this metadata for customer-facing usage displays, internal metering, debugging, and reconciliation with Console billing history.

    JSON object responses include these body fields in addition to the endpoint-specific payload:

    FieldTypeMeaning
    cost_usdnumberCost charged for this request, denominated in USD.
    usage_countintegerNumber of billable usage units applied to the request.
    servicestringName of the priced service used for the charge, such as AI Search, X Search, or Web Search.
    currencystringCurrency for cost_usd; currently USD.

    Compatible charged responses also include the same values as HTTP headers:

    HeaderMeaning
    X-Desearch-Cost-UsdCost charged for this request, in USD.
    X-Desearch-Usage-CountBillable usage units applied to the request.
    X-Desearch-ServiceName of the priced service used for the charge, same value as service.
    X-Desearch-CurrencyCurrency for the cost value; currently USD.

    Compatibility rules:

    • JSON object responses preserve their endpoint fields and add cost_usd, usage_count, service, and currency.
    • JSON arrays preserve the array body shape and expose cost metadata through the X-Desearch-* headers only.
    • text/plain responses, such as Extract output, preserve the text body and expose cost metadata through headers only.
    • Streaming/SSE responses are not buffered or reshaped; read request cost from the response headers.

    Representative charged endpoint behavior:

    EndpointSuccess body shapeCost metadata location
    POST /searchJSON objectObject body fields. $0.30 per 1,000 results, at least 10 billed ($0.003 for a default search); no results is free; the summary has no extra charge.
    POST /desearch/ai/searchJSON object for non-streaming calls; SSE for streaming callsObject body fields + headers for JSON objects; headers only for streaming/SSE.
    POST /desearch/ai/search/links/web (deprecated, use POST /search)JSON objectObject body fields + headers.
    POST /desearch/ai/search/links/twitterJSON objectObject body fields + headers.
    GET /twitterJSON array of posts, charged per unique post returned (never more than count); an empty array when no posts are found, not charged; 503 JSON object with error_code when X search is unavailable, not chargedHeaders only for arrays, including the empty array; object body fields + headers for the 503 body.
    GET /webJSON objectObject body fields + headers.
    GET /web/extracttext/plainHeaders only; text body is unchanged.
    POST /desearch/people/search/{x,youtube,linkedin,tiktok}JSON object (the created task)Object body fields + headers. They report the reservation; see People Search billing below.
    POST /desearch/companies/search, POST /desearch/companies/enrichJSON object (the created task)Object body fields + headers. They report the reservation; see Company Search billing below.
    Monitor runs (/desearch/monitors)None: runs bill in the backgroundEach run shows what it held (reserved_usd) and was billed (cost_usd); see Monitors billing below. Monitor routes are free.

    GET /web/crawl is a deprecated endpoint alias for GET /web/extract (same extract pricing tier). Prefer /web/extract for new integrations. This does not deprecate a Console plan or another API family.

    Example non-streaming JSON object response:

    json
    { "answer": "ok", "cost_usd": 0.004, "usage_count": 10, "service": "AI Search", "currency": "USD" }

    Example headers for the same request:

    http
    X-Desearch-Cost-Usd: 0.004 X-Desearch-Usage-Count: 10 X-Desearch-Service: AI Search X-Desearch-Currency: USD

    Pricing

    Pricing is usage-based and may vary by endpoint, source, volume, and account agreement. Check Pricing in Console for the latest rates available to your account.

    Common pricing groups include:

    • Search: POST /search, $0.30 per 1,000 results. At least 10 results are billed, so a default search (count 10) costs $0.003, which is $3 per 1,000 default searches. A search that returns no results is free. The optional summary has no extra charge.
    • Basic services: SERP web search, X search, and link search endpoints.
    • AI-powered services: AI Search (X posts and web with an optional summary) and AI X Posts Search.
    • Social data services: TikTok, Facebook, Instagram, YouTube, Reddit, Threads, and expanded X retrieval endpoints.
    • People Search: X Profile Search, YouTube Creator Search, LinkedIn Profile Search, and TikTok Creator Search, billed per person delivered.
    • Company Search: Company Search and Company Enrich, billed per company delivered.
    • Monitors: scheduled searches, billed per run for the searches it makes plus screening per new item.
    • High-volume or enterprise usage: custom limits, volume discounts, support/SLA, invoiced billing, or dedicated agreements.

    Cost calculation example

    If an endpoint rate is $0.15 / 1,000 usage units and the request uses 10 units, the API reports:

    text
    cost_usd = 10 * (0.15 / 1000) = 0.0015

    This is why Desearch reports fractional USD values instead of cents.

    Minimum charge for X searches

    X Search (GET /twitter) is charged for the unique posts it returns, never more than the count you asked for, with no minimum. Posts left out because they contain none of the query's words, or because of verified or blue_verified, are not charged. A search that finds no posts returns 200 with an empty list and is not charged.

    Other X (Twitter) search endpoints bill for the number of posts actually returned, with a minimum of 10 posts per request. A search that returns few posts is billed at the 10-post minimum rather than the full requested count, and a request is never billed above the count you asked for. On GET /twitter/replies and GET /twitter/replies/post, a search that finds no posts returns 200 with an empty list and is not charged; the 10-post minimum applies only to responses that return posts.

    On GET /twitter/urls, each unique post returned is charged, with no minimum. A post asked for twice is returned and charged once. URLs whose post cannot be read are left out of the list and are not charged. When no post can be read, the request returns 200 with an empty list and is not charged.

    On GET /twitter/user/posts, each page is charged for the unique posts it returns, with a minimum of 10 posts and at most 60 posts per page. A page can hold more than 20 posts. A page with no posts is not charged.

    People Search billing

    People Search (POST /desearch/people/search/{x,youtube,linkedin,tiktok}) bills per person delivered, not per request:

    • Starting a search reserves count people at the service's rate. The create response reports the reservation in cost_usd and the X-Desearch-* headers, and the task shows it as reserved_cost_usd.
    • When the search ends, you pay for the qualified people it delivered, with a minimum of 1 profile or creator per search that finds anyone, on X, YouTube, LinkedIn, and TikTok. You never pay more than the reservation.
    • The rest of the reservation is returned automatically. Once the task has ended, it shows final_cost_usd and refunded_usd.
    • Without a max_cost_usd budget, a search that finds nobody is free. A search cancelled before it starts is free, with or without a budget.
    • A People Search with a max_cost_usd budget reserves the whole budget and searches until it delivers count, spends it, or reaches another limit of the search. When it completes or is cancelled while it runs, its search effort (what the search cost, with a margin) is billed when the people delivered do not cover it, up to the budget, also when it finds nobody. A search cancelled while queued or paused, or a failed or partial search, pays only for the people it delivered. The unused part of the budget is returned.
    • Polling a task, reading results, exports, pause, resume, cancel, and the profiles library are free.

    Rates per 1,000 people are on the pricing page. See the People Search guide for the full flow.

    Company Search billing

    Company Search (POST /desearch/companies/search) and Company Enrich (POST /desearch/companies/enrich) bill per company delivered, not per request:

    • Company Search reserves count companies at its rate. Company Enrich reserves one company per domain you send. The create response reports the reservation in cost_usd and the X-Desearch-* headers, and the task shows it as reserved_cost_usd.
    • When the task ends, you pay for the companies it delivered, with a minimum of 1 company for both Company Search and Company Enrich when it delivers any. You never pay more than the reservation.
    • The rest of the reservation is returned automatically, including duplicate domains on Company Enrich. Once the task has ended, it shows final_cost_usd and refunded_usd.
    • A company left out by min_score or filters.require, or whose site cannot be read, is not billed. Without a max_cost_usd budget, a task that delivers no companies is free. A task cancelled before it starts is free, with or without a budget.
    • A Company Search with a max_cost_usd budget reserves the whole budget and searches until it delivers count or spends it. When it completes or is cancelled while it runs, its search effort (what the search cost, with a margin) is billed when the companies delivered do not cover it, up to the budget, also when it delivers no companies. A task cancelled while queued or paused, or a failed or partial task, pays only for the companies it delivered. The unused part of the budget is returned.
    • Polling a task, reading results, exports, pause, resume, cancel, and the task list are free.

    Rates per 1,000 companies are on the pricing page. See the Company Search guide for the full flow.

    Monitors billing

    Monitors (/desearch/monitors) bill per run, not per call. Creating, reading, changing, and deleting monitors is free, and so are the estimate, webhook test, and rotate-secret calls.

    • Each run pays for its searches (or the feeds of its watched accounts and hashtags) at their list price, the same price as calling those endpoints yourself, plus AI screening for each new item it screens. summary: true adds to the cost of each run.
    • TikTok, Instagram, Facebook, and Meta ads are billed per read, not per post: one read per watched account, hashtag, or Facebook page, per TikTok keyword search, and per Ad Library search, whatever the number of posts it returns. A run's hold counts two reads per TikTok account. X timelines, subreddits, and Threads profiles keep their list prices.
    • When a run is queued, its maximum cost (max_run_cost_usd on the monitor) is held on your balance. When it ends, it is billed what it used, never more than the hold, and the rest is released. The run shows reserved_usd and cost_usd.
    • When your balance runs out, the monitor pauses with paused_reason: "insufficient_credits" until you resume it.
    • POST /desearch/monitors/estimate returns the most one run and one month can cost for a set of sources, before you create the monitor.

    The screening and per-read rates are on the pricing page. See the Monitors guide for the full flow.