Sonar Deep Research API Pricing
Retired September 27, 2026: Perplexity ended Sonar Chat Completions support on that date. Synchronous sonar-deep-research requests are being reformulated as Agent API requests, rolling out by model, and asynchronous ones are no longer accepted; Perplexity's suggested starting point is the Agent API high preset. The rates below are the last the Sonar API listed, kept for invoice checks. Sonar Deep Research was the deep-research tier, with extra citation, reasoning and search-query billing. The Sonar API listed it at $2/M input and $8/M output, with request and search fees billed separately. Checked against Perplexity's migration guide.
Run the numbers.
Live calculator pre-loaded with current Sonar Deep Research token rates. Perplexity request and search fees are documented in the scenarios and FAQ, because they are separate from token pricing.
Real-world presets.
Low-context research
Medium-context research
Competitive brief
Source-backed audit
This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Sonar Deep Research Current | $2.00 | $8.00 | $2.48 agentic 92/8 | not listed | Multi-step deep research |
| Sonar (Agent API) | $0.25 cache $0.0625 | $2.50 | $0.289 −8.6× | not listed | Perplexity's per-token model on the Agent API |
| Sonar | $1.00 | $1.00 | $1.00 retired | not listed | Low-cost web-grounded answers |
| Sonar Pro | $3.00 | $15.00 | $3.96 retired | not listed | Higher-quality search answers |
| Sonar Reasoning Pro | $2.00 | $8.00 | $2.48 retired | not listed | Reasoned web research |
| Reka Flash | $0.80 | $2.00 | $0.896 −2.8× | not listed | Fast chat and production assistants |
| Aya Expanse 32B | $0.50 | $1.50 | $0.58 −4.3× | 128K | Multilingual generation and translation |
| Reka Edge | $0.10 | $0.10 | $0.10 −25× | not listed | Lowest-cost Reka chat |
Frequently asked.
Practical Sonar Deep Research pricing questions, with public list prices separated from assumptions and add-on fees.
Q · 01 What happened to Sonar Deep Research? +
sonar-deep-research the suggested starting point is the Agent API high preset. Perplexity does not say which rates a reformulated request bills at. Its own model on the Agent API, Sonar (Agent API), lists at $0.25/M input and $2.50/M output, with tools billed per call.Q · 02 What is Sonar Deep Research priced at? +
$2/M input and $8/M output on Perplexity's official pricing page, verified 2026-06-14.Q · 03 Is prompt caching priced separately? +
$2/M, instead of inventing a discount.Q · 04 How is the effective price calculated? +
$2.48/M.Q · 05 Does this include every possible fee? +
$2/M, reasoning tokens at $3/M, and search queries at $5/1K. The token calculator is therefore a base estimate, not a full invoice simulator.Q · 06 Are there free trials or production keys? +
Q · 07 Which API model ID should I use? +
sonar-deep-research belonged to the Sonar Chat Completions API, which Perplexity stopped supporting on September 27, 2026. Its suggested starting point is the Agent API high preset; Perplexity's own per-token model there is perplexity/sonar, priced on the Sonar (Agent API) page.