Tavily on x402: give an agent a current web question

On this page

We investigated Tavily's x402 search route with a familiar business job in mind: preparing a current briefing from web sources. We checked the provider documentation, fetched its public pricing record and compared the marketplace listing. Here are the request details and choices that matter when you try it.

One finding affects how you design the task: Tavily's direct x402 route documents search with optional page content, while its separate Extract API uses a different access path. We've kept those options distinct so you can specify what the agent should bring back.

Our homepage's Tavily listing identifies the provider-operated x402.tavily.com endpoint. We'll use an illustrative venue briefing to show how its search fields could serve an events team that needs recent announcements and the original pages behind them.

Ask a question with a useful time boundary

For an illustrative venue briefing, the agent could search for recent programme announcements from Perth Convention and Exhibition Centre. The output would be a short list of candidate developments with supporting links, ready for a coordinator to check against the event plan.

The endpoint is POST https://x402.tavily.com/search. This sample request is illustrative and has not been purchased or run:

{
  "query": "Perth Convention and Exhibition Centre event announcements",
  "topic": "general",
  "time_range": "month",
  "max_results": 5,
  "include_answer": false,
  "include_raw_content": true
}

Tavily's Search reference describes the query, result-count and time filters. It also supports domain inclusion and exclusion. include_answer requests a generated answer when enabled; include_raw_content requests page content alongside the search result. Those are different outputs, so choose the one the next step needs.

The sample leaves answer generation off because the agent's immediate job is gathering evidence. It asks for page content so the briefing can work from more than a short snippet. The time filter narrows discovery; it is not proof that every announcement is new or that a reported event is still going ahead.

Read the sources before writing the update

The documented response includes a results array with titles, URLs, content and relevance scores. Optional raw content gives the agent more source material to inspect. Tavily's x402 documentation says this route returns the standard search response and always runs Advanced Search.

For the illustrative venue task, ask the agent to separate the publication date from the date of the event. Two search results may repeat the same announcement. A venue page and an organiser's page may also serve different purposes: one describes the location, while the other carries the current programme.

A useful briefing could contain three fields for each item: what changed, why it matters to this event and where to verify it. Where the sources do not resolve a question, identify the question for the coordinator instead of manufacturing a definite answer.

Price and payment route

On 7 September 2026, Tavily's machine-readable pricing returned a fixed Advanced Search price of 0.01 USDC per call, on Base. The Agentic Market API entry lists the same method, endpoint and amount.

An x402 client obtains payment requirements, signs the authorisation and retries the search. Tavily documents a settlement receipt in the PAYMENT-RESPONSE header. The amount is in USDC; it is not an AUD tariff or an Agentic Economy account price.

The provider's integration instructions include client examples. Use those when moving from a proposed query to a paid call, and check the live offer for that request.

Search and extraction have different entry points

Tavily also has a separate Extract API. Its reference documents POST https://api.tavily.com/extract, taking known URLs and returning page content, with failed results identified separately. The reference shows account-token authentication.

The direct x402 documentation and pricing we checked list /search. They do not establish https://x402.tavily.com/extract as an available paid route. Requesting raw content in a search response is therefore the documented choice in this example; do not invent an extraction URL by changing the path.

If you already know the URLs and want a documented x402 content-retrieval route, examine Exa's Contents endpoint. If the task requires clicking through an interface, examine Browserbase.

We verified the public documentation, pricing JSON and listing without making a paid call. To try the venue task, begin with one question whose answer you can check, then compare the returned sources with the briefing you need to deliver. Our business use-case examples explain where that evidence-gathering step fits in a larger job.