Website Q&A widget
sonar · medium context · 10,000 requests/month · ~150 input and ~400 output tokens each
| Input tokens (1.5M × $1) | $1.50 |
| Output tokens (4M × $1) | $4.00 |
| Request fees (10K × $8/1K) | $80.00 |
| Total | $85.50 |
Short answer
From about half a cent per question with sonar
You pay for tokens in and out, plus a small fee per request for the web search. With the basic sonar model, a short question and answer typically costs less than one cent. API billing is separate from Perplexity Pro and Max subscriptions.
Basics
How Perplexity API billing works
Each Sonar call is billed in three parts.
Part 1Input tokens
The text you send: your question, system prompt and any chat history. Priced per million tokens.
Part 2Output tokens
The answer Perplexity writes back. Usually the biggest token cost, especially on sonar-pro.
Part 3Request fee
A flat fee per call for the live web search, priced per 1,000 requests and set by search context size.
cost = (input tokens × input price + output tokens × output price) ÷ 1,000,000 + requests ÷ 1,000 × request fee
What is a token?
What is Perplexity AI?
A token is a chunk of text, roughly 4 characters or ¾ of a word in English. 1,000 tokens ≈ 750 words. Exact counts vary by language and wording.
Sonar
Sonar model prices
Token prices per 1 million tokens; request fees per 1,000 requests.
| Model ID | Input | Output | Low context | Medium | High |
|---|---|---|---|---|---|
| sonar | $1 | $1 | $5 | $8 | $12 |
| sonar-pro | $3 | $15 | $6 | $10 | $14 |
| sonar-reasoning-pro | $2 | $8 | $6 | $10 | $14 |
Which to pick? Start with sonar. Use sonar-pro for harder, multi-part questions and sonar-reasoning-pro when you need step-by-step reasoning. See the Sonar API guide.
Search context
What search context size means for your bill
Search context controls how much web material Perplexity gathers for each answer. More context can mean better answers on hard questions, and a higher request fee.
Low
Cheapest and fastest. Good for simple facts and short answers.
Medium
A balance of cost and depth for most apps.
High
Most thorough. Use for research and complex questions.
Deep Research
sonar-deep-research pricing
Deep Research is billed differently because it runs many searches and reasons across many sources.
| Charge | Price | What it is |
|---|---|---|
| Input tokens | $2 / 1M | Your prompt |
| Output tokens | $8 / 1M | The written report |
| Citation tokens | $2 / 1M | Source text used to ground and cite the report |
| Reasoning tokens | $3 / 1M | The model’s internal reasoning while researching |
| Searches | $5 / 1K | Each web search it runs |
Other APIs
Search, Agent and Embeddings API pricing
| API / tool | Price | Notes |
|---|---|---|
| Search API | $5 / 1K requests | Ranked web results, no written answer |
| Search API (fast) | $1 / 1K requests | Quicker, lighter results |
| Agent API: web search | $0.0025 / call | Plus the chosen model’s token price |
| Agent API: image search | $0.0025 / call | |
| Agent API: fetch URL | $0.0005 / call | |
| Agent API: people / finance search | $0.005 / call | |
| Agent API: sandbox session | $0.03 / session | Up to 20 minutes |
| pplx-embed-v1-0.6b | $0.004 / 1M tokens | 1,024-dimension embeddings |
| pplx-embed-v1-4b | $0.03 / 1M tokens | 2,560-dimension embeddings |
The Agent API also charges each model’s own token price (models from OpenAI, Anthropic, Google, xAI and others). These vary by model; see the official pricing page.
Examples
Worked examples
Real arithmetic using the prices above. Your token counts will differ.
Customer support assistant
sonar-pro · high context · 50,000 requests/month · ~500 input and ~800 output tokens each
| Input tokens (25M × $3) | $75.00 |
| Output tokens (40M × $15) | $600.00 |
| Request fees (50K × $14/1K) | $700.00 |
| Total | $1,375.00 |
One Deep Research report
sonar-deep-research · illustrative usage: 2K input, 8K output, 20K citation, 60K reasoning tokens, 20 searches
| Input (2K × $2/1M) | $0.00 |
| Output (8K × $8/1M) | $0.06 |
| Citation tokens (20K × $2/1M) | $0.04 |
| Reasoning tokens (60K × $3/1M) | $0.18 |
| Searches (20 × $5/1K) | $0.10 |
| Total | $0.39 |
Perplexity API cost calculator
Choose a model and enter your expected usage. Everything updates as you type.
Token cost–
Request fees–
Estimated total / month–
Per request–
Save money
How to lower your Perplexity API bill
Default to sonar, low context
Route only hard questions to sonar-pro, reasoning or high context.
Keep answers short
Ask for concise answers and set a max token limit. Output tokens cost the most on sonar-pro.
Trim the prompt
Send only the chat history you need; long histories add input tokens on every call.
Cache common questions
Store recent answers for repeated queries instead of paying again.
Use the Search API for links
If you only need results, fast Search API calls cost $1 per 1,000.
Watch usage and set alerts
Track spend in the API console so a loop or bug can’t run up a large bill.
Subscriptions vs API
Perplexity Pro vs the API: not the same bill
Pro / Max subscription
- Flat monthly price ($20 Pro, $200 Max)
- For using the Perplexity apps yourself
- Usage limits, no per-token billing
API
- Pay-as-you-go credit in the API console
- For building Perplexity into your own software
- Billed per token, request and search
See Perplexity subscription pricing for app plans.
FAQ
Perplexity API pricing questions
How much does the Perplexity API cost?
It’s pay-as-you-go. The basic sonar model costs $1 per million input tokens and $1 per million output tokens, plus a request fee of $5–$12 per 1,000 requests depending on search context. A typical short Q&A call with sonar costs well under one cent.
Is there a free tier for the Perplexity API?
There isn’t an ongoing free API tier. You add credit in the API console and pay for what you use. Some Perplexity Pro subscriptions have included a small monthly API credit in the past; check your account, as this has changed over time.
What is the request fee?
Every Sonar call pays a small fee for the web search behind it, charged per 1,000 requests. It depends on the model and the search context size you choose (low, medium or high).
Does Perplexity Pro include API access?
No. Pro and Max subscriptions cover the Perplexity apps. API usage is billed separately through the API console.
What is the cheapest way to use the Perplexity API?
Use sonar with low search context, keep prompts and answers short, cache repeated questions, and only route hard questions to sonar-pro or reasoning models. Use the Search API at $1 per 1,000 in fast mode if you only need links.
Why is Deep Research priced differently?
Deep Research runs many searches and reasons over many sources, so it bills citation and reasoning tokens and charges per search ($5 per 1,000) instead of a flat request fee. A single report usually costs well under a dollar, but varies with the topic.
Where can I check current prices?
Perplexity’s official API pricing page in the docs. Prices on this page were checked on 6 October 2026.
More API guides
API starting guideFirst call in 5 steps
Get an API keyConsole setup
Sonar APIModels and parameters
Search APIRaw results
Python guideSDK examples
TroubleshootingErrors and fixes
Source: Perplexity API pricing (official docs), checked 6 October 2026.
Start building
Add credit and make your first call
Top up a small amount in the API console, try sonar, and watch your real costs in the usage dashboard.