Locus product guide · Updated November 10, 2026

Perplexity API pricing: Search, Agent API and Sonar costs

Perplexity’s Search API costs $5 per 1,000 requests, or $1 per 1,000 with Fast Search, and one request can carry up to 5 queries. The Agent API bills each model’s token rates plus tool fees, such as $0.0025 per web search. The perplexity/sonar model costs $0.25 per million input tokens and $2.50 per million output tokens. Sonar Chat Completions support ended on September 27, 2026. Prices checked November 10, 2026.

How is the Perplexity API priced?

Each Perplexity API bills a different way. Search is per request. The Agent API and embeddings are per token. Rates below come from Perplexity’s pricing docs. Prices checked November 10, 2026.

How is the Perplexity API priced?
API	Billing unit	Price
Search API	Per successful request	$5 per 1,000, or $1 with Fast Search
Agent API	Per token, plus tool fees	Each model’s rate, plus $0.0025 per web search
Embeddings API	Per token	$0.004 to $0.05 per million tokens
Decisions API	Per input token	$0.01 or $0.02 per million, plus $0.001 per grounding search
Orchestrator API	Per token, in private preview	Model rates with no request fee

What do Agent API tools and models cost?

An Agent API request costs its tokens at the chosen model’s rates plus a fee for every tool call. Perplexity’s own model, perplexity/sonar, costs $0.25 per million input tokens and $2.50 per million output tokens. The fast, low and medium effort levels run on GPT-6 Luna, listed at $0.10 input and $0.50 output per million.

Some models accept a flex tier at half the token price or a priority tier at double. The fast effort level uses priority. Tool fees do not change with the tier.

What do Agent API tools and models cost?
Agent API tool	Price per call	Notes
web_search	$0.0025	$0.001 with Fast Search
image_search	$0.0025 per successful call	$2.50 per 1,000
fetch_url	$0.0005	Fetches and extracts one URL
people_search	$0.005	Looks up professionals and employees
finance_search	$0.005	Market and financial data
sandbox	$0.03 per session	20-minute billing window, plus $0.0025 per search inside it

What happened to Sonar API pricing?

Sonar Chat Completions support ended on September 27, 2026. Synchronous and streaming Sonar requests still work: Perplexity is turning them into Agent API requests, model by model. Asynchronous Sonar requests are no longer supported and must move to background mode.

Perplexity maps each Sonar model to an Agent API effort level. The cost now follows that effort level’s model, tokens and tool calls.

What happened to Sonar API pricing?
Old Sonar model	Agent API effort level	What runs it today
Sonar	fast	GPT-6 Luna at priority, one Fast Search
Sonar Pro	fast effort	Same as Sonar
Sonar Reasoning Pro	low	GPT-6 Luna, up to 5 steps
Sonar Deep Research	high	GPT-6.1 Terra, up to 40 steps

What does the Perplexity Search API cost at 1,000 and 100,000 searches?

1,000 standard searches cost $5. 100,000 cost $500, or $100 with Fast Search. Batching cuts that further: a request with up to 5 queries is one billing unit, so 100,000 queries sent 5 at a time cost $100, or $20 with Fast Search.

For a sense of Agent API costs, Perplexity’s docs show a GPT-5.6 Sol answer with 6,599 input tokens, 657 output tokens and one web search that cost $0.029.

What does the Perplexity Search API cost at 1,000 and 100,000 searches?
Monthly workload	Standard search	Fast Search
1,000 searches	$5	$1
100,000 searches	$500	$100
100,000 queries sent 5 per request	$100	$20

What are the catches in Perplexity API pricing?

Most surprises come from rate limits and billing units, not the list prices.

The pricing page lists no free allowance. You buy credits before you call.
Agent API rate limits follow lifetime credit purchases. A new account gets 1 request a second, $50 of purchases unlocks 3, and $1,000 unlocks 33.
Polling a running Agent API job counts against the same request limit.
Token limits per model can return a 429 even when you are under your request limit.
A successful search is billed even when it returns no results. Invalid, rate-limited and failed requests are free.
Search allows 50 query units a second. Each query in a batch uses 1 unit, even though the batch bills once.

Which Perplexity endpoints does Locus Pro offer?

Locus Pro lists Perplexity search, Sonar chat and embeddings as known-vendor endpoints. Your agent pays from one prepaid credit balance with no Perplexity key. Catalog checked November 10, 2026.

Sonar chat accepts sonar, sonar-pro and sonar-reasoning-pro, priced from tokens. Locus asks for max_tokens so it can cap the charge. Deep Research is excluded. Fast Search and the full Agent API, with its third-party models and tools, are not in the Locus catalog.

Locus Pro fits agents that call Perplexity now and then next to other paid tools. One balance covers search, scraping and enrichment, each call shows its price first, and you skip a separate Perplexity account. At steady search volume, searches can cost less per call with your own Perplexity account and API key.

Which Perplexity endpoints does Locus Pro offer?
Locus endpoint	What it returns	How it is priced
Perplexity Search API	Ranked web results for up to 5 queries	Per call
Sonar chat	A cited answer from a Sonar model	From tokens, capped by max_tokens
Embeddings	Vectors for text	By tokens
Contextualized embeddings	Vectors that keep document context	By input tokens

Which Perplexity API should you use?

Choose by what you want back.

Ranked results for your own model: Search API, with Fast Search when speed and price matter more than depth.
A cited answer: the Agent API at the fast or low effort level.
Multi-step research: the Agent API at medium or high effort.
Old Sonar code: keep it running for now, and plan the move to the Agent API.
A few Perplexity calls among many tools on one bill: Locus Pro.

Frequently asked questions

Can I still call Perplexity Sonar models?

Yes, for now. Sonar Chat Completions support ended on September 27, 2026, but synchronous and streaming requests are being turned into Agent API requests. Asynchronous Sonar requests no longer work.

How can I lower my Perplexity Search API bill?

Use Fast Search at $1 per 1,000 requests, and send up to 5 queries in one request. A batch of 5 bills as a single request.

Does Perplexity charge for searches that return nothing?

Yes. A successful search request is billed even with no results. Invalid requests, rate-limited requests and upstream failures are not billed.

How do I get higher Perplexity API rate limits?

Buy more credits over time. Usage tiers follow total lifetime purchases, from Tier 0 at $0 to Tier 5 at $5,000, and you keep a tier once you reach it.

Implementation references

Use these first-party references for current request contracts and account requirements. Tool availability, prices, and negotiated terms can change.

Sources

Third-party details were checked on November 10, 2026. Vendors change pricing and features often, so confirm on their site before you decide.