Skip to main content
This page shows pricing information to help you understand API costs.For billing setup, payment methods, and usage monitoring, visit the Admin section. For rate limits, see the Rate Limits & Usage Tiers page.

Estimate your cost

Orchestrator API Pricing

Orchestrator API usage is billed per token at each model’s published rates, with no per-request fees. Every model has its own input and output rate, with cache reads billed at the model’s cache-read rate and reasoning tokens billed at the output rate. Some models also charge for cache writes; see Orchestrator billing. You are billed at the requested model’s rates regardless of which deployment serves the request; with perplexity/auto, at the rates of the model that served it. See Orchestrator Models & Pricing for the full per-model rate card, including cache-read rates and usage tiers. For supported OpenAI models, service tiers scale token rates by 0.5× for flex and 2× for priority. The listed rates for google/gemini-3.7-flash and google/gemini-3.8-flash apply through December 31, 2026.

Agent API Pricing

The Agent API charges for model tokens and tool calls. The catalog includes models from OpenAI, Anthropic, Google, xAI, Z.AI, DeepSeek, Moonshot AI, NVIDIA, and Perplexity.

Model Pricing

View Agent API Model Pricing

Compare model rates, cache rates, and supported service tiers.
The flex and priority service tiers scale token rates by 0.5× and 2×, respectively, on models that list those tiers. The introductory rates for google/gemini-4-argon and google/gemini-4-carbon apply through November 30, 2026. The listed rates for google/gemini-3.7-flash and google/gemini-3.8-flash apply through December 31, 2026.

Tool Pricing

When using tools with the Agent API:
Most tool costs are per invocation. sandbox is billed per started 20-minute block of container time, plus each SDK web search made inside it. Tool costs are separate from model token costs.

Search API Pricing

Billing unit: Search API charges for each successful POST /search request, not for each query in the request. A successful request containing an array of up to five queries is one billing unit. Invalid requests, rate-limited requests, and upstream failures are not billed. A successful response is billed even when it returns no results. There are no additional token-based charges.
Search API cost = successful requests ÷ 1,000 × the price per 1K requests. What a result contains, the request controls, and the limits on queries and results are in How Search Works.

Monitor API Pricing

Cost = (succeeded Search runs + counted fetched URLs from succeeded Page runs) × $0.001. Failed and skipped runs are free. A succeeded Search run is billed once, even if it performs multiple searches or returns no findings. In the calculator, choose Monitor API and enter both totals for the same period. The run’s usage.cost reports list-price cost, not the invoice or amount charged.

Rate limits

See Monitor API limits for HTTP request limits by usage tier.

Context API Pricing

The Context API returns the content of web pages you already have URLs for. It costs $1.00 per 1,000 fetched URLs ($0.001 per URL). Requests without criteria have no token billing.
Billing unit: each URL in a successful POST /v1/context response whose status.type is ok or not_modified, whatever formats asks for. Highlights and metadata-only requests with "formats": [] are billed the same way. URLs that fail are free, including invalid, blocked, empty, and timed-out URLs.
Context API cost without criteria = billable URLs (ok or not_modified) ÷ 1,000 × $1.00. With criteria, scoring is billed as Decisions API input tokens on top of the per-URL price: the page text plus your questions, for each scored page. The response reports the count as usage.criteria.input_tokens. To estimate your cost, open the calculator and enter the URLs you expect to fetch per month. Each successful response reports the billed URL count in usage.fetched_urls and its list-price cost in usage.cost.total_cost, before any account-specific discounts, credits, or adjustments.

Embeddings API Pricing

Generate high-quality text embeddings for semantic search, retrieval-augmented generation (RAG), and other machine learning applications.

Contextualized Embeddings v2

Text inputs use the token price. Image inputs use the image price.

Standard Embeddings

Contextualized Embeddings

View Embeddings API Documentation

Learn how to use the Embeddings API for semantic search, RAG, and more.
Embeddings cost = tokens ÷ 1,000,000 × the model’s rate. For example, 40M tokens on pplx-embed-context-v1-4b cost 40 × $0.05 = $2.00.

Decisions API Pricing

The Decisions API answers yes/no, multiple-choice, and scored questions about text and images with calibrated probabilities, and finds matching text spans with locate questions.
Aliases: pplx-decider-latest and perplexity/pplx-decider-latest select pplx-decider-v1.3. pplx-decider-flash-latest and perplexity/pplx-decider-flash-latest select pplx-decider-flash-v1.3. An alias bills at the rate of the model it selects. Requests that name pplx-decider-v1.1-27b or pplx-decider-v1-27b select pplx-decider-v1.2 and bill at $0.02.Billing unit: you pay for input tokens, plus web searches and URL fetches if you set mode: "web_grounding". The state, every question, any images, fetched page text, and any search results added to the state count as input; the response reports the exact number as usage.input_tokens, so you can compute the token cost of a request from the response you already receive. A request with 600 input tokens costs $0.000012 on pplx-decider-v1.3 or pplx-decider-v1.2 and $0.000006 on pplx-decider-flash-v1.3 or pplx-decider-flash-v1.2.

Web grounding

A request with mode: "web_grounding" runs one Fast Search for each of its questions, up to 100, before the model answers. Each search costs $0.001 ($1 per 1,000), in addition to input tokens, so at most $0.10 in searches per request. Questions that produce the same search query share one search. When state is a page URL, the API also fetches the page before searching. Each successful URL fetch costs $0.0005; a failed fetch is not billed. One successful fetch and three successful searches cost $0.0035, plus input tokens.
  • Billed: every search that returns, even with no results. Completed searches and successful URL fetches remain billable if the request later fails, for example because the input exceeds the token limit.
  • Not billed: a search that fails or runs out of time, a question with no text to search, and requests rejected before search: an invalid mode, more than 100 questions, a missing or unknown model, or a state that is not a string, object, or array. A request can be billed for fewer searches than it has questions.
  • Input tokens: fetched page text and search evidence are billed as input tokens and included in usage.input_tokens.
See Web grounding.

View Decisions API Documentation

Learn how to ask structured questions and read the probabilities the Decisions API returns.

Input Tokens

The number of tokens in your prompt or message to the API. This includes:
  • Your question or instruction
  • Any context or examples you provide
  • System messages and formatting
  • Tool results that the API passes to the model
Example: “What is the weather in New York?” = ~8 input tokens

Output Tokens

The number of tokens in the API’s response. This includes:
  • The generated answer or content
  • Any explanations or additional context
Example: “The weather in New York is currently sunny with a temperature of 72°F.” = ~15 output tokens
Token Calculation: 1 token ≈ 4 characters in English text. The exact count may vary based on language and content complexity.

Cost Examples

Agent API Web Search

openai/gpt-5.6-terra • 500 fresh input + 200 output tokens • 1 standard web search

Agent API Research Effort Level

low effort level example • openai/gpt-6-luna • 2,000 fresh input + 1,000 output tokens • 1 Fast Search + 1 fetch
Actual effort level costs vary with the model, token usage, and tool calls. When present, usage.cost.total_cost reports an estimate. Check Billing in the API Console for charged amounts.

Purchase Options

Perplexity API Platform on AWS Marketplace

Purchase API credits through AWS Marketplace with consolidated billing and enterprise procurement.

Contact Sales Team

Fill out our enterprise inquiry form to discuss custom pricing, dedicated support, and enterprise features for teams and organizations.