Perplexity
Perplexity Sonar API pricing guide
Understand Sonar token charges, request fees and Deep Research extras. Perplexity currently has no eligible offers in this calculator; use this guide as a billing reference and confirm current rates with Perplexity.
Updated
Read the Sonar pricing guideCompare eligible offers from other providers
Related: Cheapest Llm Api · Methodology
Understanding Perplexity Sonar API costs
Snapshot validation: . These are the calculator’s dated rates, not a new price verification. Confirm current rates and availability with the provider.
What this estimate excludes
Sonar, Sonar Pro and Sonar Reasoning Pro combine token charges with request fees that depend on search context size. Sonar Pro standard and Pro Search modes have different request fees.
Sonar Deep Research has additional billing dimensions, including citation tokens, searches and reasoning tokens. A simple input/output estimate does not represent its full invoice.
These notes cover the tracked Sonar models. Perplexity Search, Agent and Router APIs have different billing structures. The snapshot includes route-specific cache rates that are not confirmed for the Sonar API. Keep cached input at 0% for a Sonar estimate; do not apply Agent API cache rates to Sonar.
Billing documentation checked September 23, 2026: Official Perplexity pricing.
Common Sonar pricing questions
How much does Sonar Pro cost?
The snapshot lists Sonar Pro at $3 per million input tokens and $15 per million output tokens. For 1,000 requests with 1,000 uncached input and 250 output tokens each, token charges are $6.75. Request fees are additional and depend on context size and search mode. Standard and Pro Search request charges differ. See the official Sonar Pro fee tables.
This snapshot rate is not cross-checked; verify it with Perplexity before use.
Why is my bill higher than the token estimate?
The calculator estimates input/output token charges. Sonar request fees and Deep Research search, citation and reasoning charges are separate. Conversation history, longer replies and retries can also increase billable usage beyond your assumptions. Compare your endpoint, search mode and recorded usage against Perplexity’s billing dimensions; keep cache at 0% unless that endpoint explicitly supports a discount.
Tracked token rates
| Model / tier | Input / MTok | Cached / MTok | Output / MTok | Evidence |
|---|---|---|---|---|
| Sonar | $1 | Not confirmed for Sonar API | $1 | Provider source Not cross-checked Reviewed not established Token rates verified, but mandatory request fees depend on search context: $5/$8/$12 per 1K requests for Sonar, $6/$10/$14 for Sonar Pro and Reasoning Pro. Pro Search has separate higher fees. Excluded until the UI accounts for the selected search mode. |
| Sonar Pro | $3 | Not confirmed for Sonar API | $15 | Provider source Not cross-checked Reviewed not established Token rates verified, but mandatory request fees depend on search context: $5/$8/$12 per 1K requests for Sonar, $6/$10/$14 for Sonar Pro and Reasoning Pro. Pro Search has separate higher fees. Excluded until the UI accounts for the selected search mode. |
| Sonar Reasoning Pro | $2 | Not confirmed for Sonar API | $8 | Provider source Not cross-checked Reviewed not established Token rates verified, but mandatory request fees depend on search context: $5/$8/$12 per 1K requests for Sonar, $6/$10/$14 for Sonar Pro and Reasoning Pro. Pro Search has separate higher fees. Excluded until the UI accounts for the selected search mode. |
| Sonar Deep Research | $2 | Not confirmed for Sonar API | $8 | Provider source Not cross-checked Reviewed not established Requires separate citation ($2/M), reasoning ($3/M) and search-query ($5/1K) usage in addition to $2/M input and $8/M output. Excluded from token-only comparisons until those billing dimensions are supported. |