How is Perplexity API different from ChatGPT or Claude API?

Perplexity API is search-augmented - each API call runs a web search and includes the results as context for the model. This means answers are grounded in real-time information with citations. Input tokens are higher (you pay for the search context) but you get verified, up-to-date answers without needing a separate RAG pipeline.

What is the difference between Sonar and Sonar Pro?

Sonar is the standard model for search-augmented generation with basic citations ($1.00/M input, $5.00/M output). Sonar Pro ($3.00/M input, $15.00/M output) provides deeper research with more sources, longer-form answers, and better reasoning. Use Sonar for simple Q&A and Sonar Pro for in-depth research and analysis.

Why are input tokens higher for Perplexity API?

Perplexity's search context adds approximately 500-2000 tokens per call depending on the query complexity and number of search results. This context is included in your input token count. Factor this in when budgeting - a 1000-token prompt may actually cost as 2000-3000 input tokens after search augmentation.

Can I control the number of search results?

Yes, Perplexity's API allows you to configure search parameters including the number of sources, search domain filters, and recency requirements. Fewer search results means lower input token costs but potentially less thorough answers. Balance thoroughness with cost based on your use case.

Does Perplexity offer volume discounts?

Perplexity offers custom pricing for high-volume users (typically $5K+/month). Contact their sales team for committed usage tiers. They also offer a free tier with limited credits for prototyping and evaluation. Enterprise plans include dedicated support and custom rate limits.

What are the best use cases for Perplexity API?

Perplexity API excels at: research assistants that need cited answers, customer support systems that need current product information, knowledge bases that update automatically, fact-checking and verification tools, competitive analysis, and any application where accuracy and source attribution are critical.

How does Perplexity handle citations and source attribution?

Perplexity returns citations inline in the response and provides a sources object with URLs and metadata. This makes it easy to display verified answers with clickable sources. Citations are formatted similarly to the Perplexity chat interface users are familiar with.

Can I use Perplexity API without web search?

Perplexity API is designed for search-augmented generation. If you don't need web search capabilities, a standard LLM API (Claude, GPT, Gemini) will be more cost-effective. Perplexity's value proposition is the integration of search and generation in a single API call with no additional infrastructure.

Perplexity API費用計算機

SonarおよびSonar ProモデルのPerplexity API費用をトークン数と使用量で計算します。