Perplexity

Get a Perplexity API key

Usage

.env

PERPLEXITY_API_KEY=

import { TokenJS } from 'token.js'

// Create the Token.js client
const tokenjs = new TokenJS()

async function main() {
  // Create a model response
  const completion = await tokenjs.chat.completions.create({
    // Specify the provider and model
    provider: 'perplexity',
    model: 'llama-3-70b-instruct',
    // Define your message
    messages: [
      {
        role: 'user',
        content: 'Hello!',
      },
    ],
  })
  console.log(completion.choices[0])
}
main()

Supported Models

Model	Chat Completion	Streaming	JSON Output	Image Input	Function Calling	N > 1
llama-3-sonar-small-32k-chat	✅	✅	➖	➖	➖	➖
llama-3-sonar-small-32k-online	✅	✅	➖	➖	➖	➖
llama-3-sonar-large-32k-chat	✅	✅	➖	➖	➖	➖
llama-3-sonar-large-32k-online	✅	✅	➖	➖	➖	➖
llama-3-8b-instruct	✅	✅	➖	➖	➖	➖
llama-3-70b-instruct	✅	✅	➖	➖	➖	➖
mixtral-8x7b-instruct	✅	✅	➖	➖	➖	➖

Legend

Symbol	Description
✅	Supported by Token.js
➖	Not supported by the LLM provider, so Token.js cannot support it

Additional Resources

• Perplexity Documentation