Sonar Deep Research
FrontierPerplexity · Released March 2025
Sonar Deep Research is a research-focused model designed for multi-step retrieval, synthesis, and reasoning across complex topics. It autonomously searches, reads, and evaluates sources, refining its approach as it gathers...
It is part of Perplexity's Sonar line, built specifically for search-grounded, citation-backed answers.
Facts and pricing
- Provider
- Perplexity
- Context window
- 128K tokens
- Input price
- 22.0 credits / 1M tokens ($2.00 raw)
- Output price
- 88.0 credits / 1M tokens ($8.00 raw)
- Vision (image input)
- No
- Tool calling
- No
- Extended reasoning
- No
Credits are what Zeplik bills: 1 credit = $0.10, computed from the raw provider rate with a 1.10x margin. Raw prices shown per 1M tokens.
Try Sonar Deep Research now
What Sonar Deep Research is best for
- Hard problems where answer quality matters more than cost: strategy, analysis, difficult writing
Example prompts
Prompts that suit a frontier model like Sonar Deep Research:
- Review this business plan and give me the three hardest questions an investor would ask
- Rewrite this announcement so it lands with both engineers and executives
- I am choosing between two job offers, help me think through the tradeoffs
- Draft a technical design doc for a rate limiter, including failure modes
Perplexity family
| Model | Context | Input cr/M | Output cr/M | Released |
|---|---|---|---|---|
| Sonar Pro Search | 200K | 33.0 | 165 | October 2025 |
| Sonar Reasoning Pro | 128K | 22.0 | 88.0 | March 2025 |
| Sonar Pro | 200K | 33.0 | 165 | March 2025 |
| Sonar Deep Research | 128K | 22.0 | 88.0 | March 2025 |
| Sonar | 127K | 11.0 | 11.0 | January 2025 |
Frequently asked questions
- What is Sonar Deep Research?
- Sonar Deep Research is an AI model by Perplexity. Sonar Deep Research is a research-focused model designed for multi-step retrieval, synthesis, and reasoning across complex topics. It autonomously searches, reads, and evaluates sources, refining its approach as it gathers...
- How much does Sonar Deep Research cost on Zeplik?
- Input tokens cost 22.0 credits per million and output tokens 88.0 credits per million (1 credit = $0.10; the raw provider rates are $2.00 and $8.00 per million). New accounts start with free credits.
- How long can a conversation with Sonar Deep Research be?
- Sonar Deep Research has a 128K-token context window (128,000 tokens), which covers the conversation plus any documents you attach.
- Does Sonar Deep Research support images and tools?
- Sonar Deep Research is text-only and does not support tool calling.
Related models
Exclusively available on the OpenRouter API, Sonar Pro's new Pro Search mode is Perplexity's most advanced agentic search system. It is designed for deeper reasoning and analysis. Pricing is based...
Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) Sonar Reasoning Pro is a premier reasoning model powered by DeepSeek R1 with Chain of Thought (CoT). Designed for...
Note: Sonar Pro pricing includes Perplexity search pricing. See [details here](https://docs.perplexity.ai/guides/pricing#detailed-pricing-breakdown-for-sonar-reasoning-pro-and-sonar-pro) For enterprises seeking more advanced capabilities, the Sonar Pro API can handle in-depth, multi-step queries with added extensibility, like...
Sonar is lightweight, affordable, fast, and simple to use — now featuring citations and the ability to customize sources. It is designed for companies seeking to integrate lightweight question-and-answer features...
GPT-5.6 Luna Pro is the same underlying model as [GPT-5.6 Luna](https://openrouter.ai/openai/gpt-5.6-luna), served with `reasoning.mode` set to `pro` for higher-quality responses on complex tasks. Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode
GPT-5.6 Luna is a fast, cost-efficient model in OpenAI's GPT-5.6 series. It is suited for high-volume, latency-sensitive tasks such as chat, classification, and lightweight agentic workflows, providing capable reasoning for...