About Perplexity

Perplexity AI is a conversational search engine that synthesizes answers from real-time web sources and cites them inline. Users can ask follow-up questions, switch between frontier AI models (GPT-5.4, Claude Sonnet 4.6, Claude Opus 4.8, Gemini 3.1 Pro, and others on paid tiers), and conduct deep research. It offers individual plans from free to $200/month (Max), enterprise plans from $40 to $325 per user/month, and a developer Sonar API. Used by individuals, researchers, and enterprises for knowledge work and competitive intelligence.

AI Agent Use Cases

- Real-time web search with inline citations for every answer

- Multi-model switching (GPT-5.4, Claude, Gemini, Sonar) on Pro and above

- Model Council: parallel three-model comparison on Max tier

- Sonar developer API for programmatic search-augmented responses

- HIPAA and SOC 2 Type II compliance on Enterprise plans

- 10,000 monthly Computer credits and Labs access on Max tier

Available Actions

These are the specific actions that AI agents can perform with this tool

Chat Completions

15 inputs

This endpoint generates a completion for a given conversation using a specified model.

Inputs

model

The name of the model that will complete your prompt.

messages

A list of messages representing the conversation so far.

maxTokens

The maximum number of completion tokens returned by the API.

temperature

The amount of randomness in the response, valued between 0 and 2.

topP

The nucleus sampling threshold, valued between 0 and 1.

searchDomainFilter

A list of domains to limit search results to.

returnImages

Specifies if search results should include images.

returnRelatedQuestions

Specifies if related questions should be returned.

searchRecencyFilter

Filters search results based on a specific time frame.

topK

The number of tokens to keep for top-k filtering.

stream

Determines if the response is streamed incrementally.

presencePenalty

Positive values increase the likelihood of discussing new topics.

frequencyPenalty

Decreases likelihood of repetition based on prior frequency.

responseFormat

Enables structured JSON output formatting.

webSearchOptions

Configuration for using web search in model responses.