Perplexity
5 models built · 5 served
Perplexity AI, Inc. is a US company founded in 2022 that builds an AI answer engine, and the Sonar models available through OpenKey are the API version of that product: LLM responses grounded in cited, live web sources rather than training-data recall alone. All five Sonar models share this search-first design instead of splitting into separate chat/code/vision product lines like most labs. Pricing runs from $1/$1 per million input/output tokens on base Sonar up to $3/$15 on the Pro and Pro Search tiers, with two mid-tier reasoning/research variants at $2/$8. Context windows top out at 200,000 tokens on the Pro models. None of the five are free. On OpenKey, provider list price gets a flat 3% fee added, so Sonar's $1 input becomes $1.03 (1 x 1.03).
Lineup
Start with Sonar ($1/$1 per million tokens, 127,072 context) for straightforward questions that need current web citations without much reasoning depth — it's the cheapest way in. Step up to Sonar Reasoning Pro or Sonar Deep Research (both $2/$8, 128,000 context) when you need multi-step reasoning or research synthesis over search results. For the largest context (200,000 tokens) and multimodal input (text+image), use Sonar Pro or the newest model, Sonar Pro Search (released 2025-10-30), both priced at $3/$15 per million tokens — the top end of the lineup. There's no free tier here, so pick based on how much reasoning and context you actually need rather than defaulting to the priciest option.
- Sonar Pro SearchOct 30, 2025 200K $3.09 $15.45 ctx 200K · in $3.09 · out $15.45
- Sonar Deep ResearchMar 7, 2025 128K $2.06 $8.24 ctx 128K · in $2.06 · out $8.24
- Sonar ProMar 7, 2025 200K $3.09 $15.45 ctx 200K · in $3.09 · out $15.45
- Sonar Reasoning ProMar 7, 2025 128K $2.06 $8.24 ctx 128K · in $2.06 · out $8.24
- SonarJan 27, 2025 127K $1.03 $1.03 ctx 127K · in $1.03 · out $1.03
Prices per 1M tokens, flat 3% fee included.
Models Perplexity serves
Beyond its own models, Perplexity hosts 5 models in the OpenKey catalog. Prices are Perplexity’s own list price per 1M input tokens, before our fee.
| Model | Built by | Context | Input /M |
|---|---|---|---|
| Sonar | Perplexity | 127K | $1.00 |
| Sonar Deep Research | Perplexity | 128K | $2.00 |
| Sonar Pro | Perplexity | 200K | $3.00 |
| Sonar Pro Search | Perplexity | 200K | $3.00 |
| Sonar Reasoning Pro | Perplexity | 128K | $2.00 |
From the OpenRouter catalog, Jul 28, 2026. All providers
Status page ↗Privacy policy ↗Terms of service ↗
Questions
- What's the price range for Perplexity's Sonar models on OpenKey?
- Provider list prices run from $1/$1 per million input/output tokens (Sonar) up to $3/$15 (Sonar Pro and Sonar Pro Search). OpenKey adds a flat 3% fee on top of provider price, so Sonar becomes $1.03/$1.03 and Sonar Pro becomes $3.09/$15.45 per million tokens.
- Which Perplexity model should I use for cheap, simple lookups?
- Sonar, at $1/$1 per million tokens with a 127,072-token context window, is the cheapest of the five Sonar models and the right default for basic web-grounded Q&A that doesn't need heavy reasoning or a huge context window.
- Does Perplexity offer any free models on OpenKey?
- No. All five Sonar models — Sonar, Sonar Pro, Sonar Pro Search, Sonar Deep Research, and Sonar Reasoning Pro — are paid, with input pricing starting at $1 per million tokens on the base Sonar model.