Tavily

Tavily provides search APIs for agent and research workflows. The calculator converts monthly searches into the verified credit-based billing rule for the catalogued profile.

Categories
Search
Models and endpoints
1
Guide reviewed
Sep 24, 2026

Models, translated into money

Choose a budget and compare what each model delivers. Every estimate includes its assumptions and pricing sources.

Budget · USD
Compare1 model
Budget USD

Tavily Basic Search

Expired

Pricing is unavailable for this model.

How we calculated thisPricing unavailable
UnitOne search

Official pricing

tavily-basic-payg · Current · Standard billing

Metered
$0 fixed · 0 included · $0.008 per 1

Verified Oct 8, 2026

Model details

Model ID
POST /search (search_depth=basic, no paid content extraction)
search Depth
basic
max Results
5
include Raw Content
No

Catalog 2026-10-08-7b05c4ce

About Tavily

Coverage, billing details, and what to check before you build.

Tavily is represented as a search provider for applications that need web results inside research or agent workflows. The calculator does not assume that every chat message requires a search. Search demand is enabled through the Research scenario or entered directly as an exact monthly count.

Where Tavily fits

A search API discovers relevant pages and returns result metadata. It is not the same billing category as downloading and parsing full pages. A research task can therefore create language-model calls, Tavily searches, and separate scraping operations. Keeping those quantities separate makes it possible to compare search providers without changing the rest of the stack.

The current catalog models one specific search profile. Options that consume extra credits or trigger paid content extraction are not silently included. If the planned application uses a deeper search mode, more results, or additional content operations, the workload must be represented by a matching verified offer before the estimate can claim coverage.

How search volume becomes cost

In scenario mode, monthly searches are derived from users, research tasks per user, and searches per task. The derived task count is rounded once before the per-task multiplier is applied. In exact mode, the supplied monthly search count is the only source of demand.

The plan card above shows the current billing rule, included units or fixed fee when applicable, effective rate, verification date, and official sources. These numbers come from the versioned pricing catalog. This article avoids copying them so that a rate update cannot leave conflicting prices on the same page.

What the estimate does not cover

Search result quality, latency, and coverage vary by query and cannot be inferred from price. Fastest and Balanced recommendations require a recent comparable evidence cohort. When such evidence is absent, the calculator continues to compare exact cost but does not manufacture a performance ranking.

The budget also excludes taxes, application hosting, retries, storage, and downstream model calls. Search results may lead to scraping or extraction charges from another provider. Those appear as separate components when the Research scenario includes pages per task.

Estimate with Tavily

Use the active offer above to open the calculator with Research enabled and Tavily locked for search. The language-model and scraping components remain unlocked, so the total still reflects a complete compatible stack. Adjust tasks per user and searches per task to match the product rather than relying on the defaults.

Review the displayed verification date and official pricing documentation before procurement. The calculator is a planning tool, not a quote from the provider.