Perplexity Block

The Perplexity block generates completions using Perplexity AI models with real-time knowledge and search capabilities. Reach for it when you need responses, answers, or generated content that benefit from up-to-date web search grounding.

Overview

When to Use

• Answer questions that require current, real-time information from the web.
• Generate content or summaries grounded in live search rather than static training data.
• Run quick research-style prompts where Perplexity's Sonar models add freshness.
• Drive a conversational response with a custom system prompt to steer tone and behavior.
• Tune output determinism and length via temperature and max tokens for predictable results.

Configuration

System Prompt

Optional instructions that guide the model's behavior and persona.

User Prompt

The required prompt content sent to the model.

Model

Dropdown selecting the model to use. Options: Sonar (sonar, default), Mistral (mistral), Claude-3-Opus (claude-3-opus), Claude-3-Sonnet (claude-3-sonnet), Command-R (command-r), and GPT-4o (gpt-4o).

Temperature

Slider (0–1, default 0.7) controlling response randomness.

Max Tokens

Optional cap on the number of tokens generated in the response.

API Key

Your required Perplexity API key (stored securely).

Inputs & Outputs

• Inputs:

  • content (string) — User prompt content sent to the model (required)
  • systemPrompt (string) — System instructions to guide model behavior (optional)
  • model (string) — AI model to use (e.g. sonar)
  • max_tokens (string) — Maximum output tokens (optional)
  • temperature (string) — Response randomness, 0–1 (optional)
  • apiKey (string) — Perplexity API key (required)

• Outputs:

  • content (string) — Generated response text
  • model (string) — Model identifier used for the request
  • usage (json) — Token usage breakdown (prompt_tokens, completion_tokens, total_tokens)

Tools

Perplexity Chat (perplexity_chat) — Posts a chat-completions request to https://api.perplexity.ai/chat/completions. Builds a messages array (optional system message + user message), passes the chosen model and optional max_tokens/temperature, and returns the generated text along with model name and token-usage stats.

YAML Example

perplexity_1:
  type: perplexity
  name: "Perplexity"
  inputs:
    systemPrompt: "You are a concise research assistant."
    content: "{{start.input}}"
    model: "sonar"
    temperature: 0.7
    max_tokens: 1024
    apiKey: "{{PERPLEXITY_API_KEY}}"
  connections:
    outgoing:
      - target: next-block-id