GitHub Docs

Perplexity Provider

Perplexity combines LLM reasoning with real-time web search, producing responses grounded in current information. Their Sonar models are uniquely suited for research-heavy coding tasks.

Get an API Key

  1. Go to perplexity.ai/settings/api
  2. Create an account or sign in
  3. Generate an API key and copy it

Configure VibeCody

Option 1: Environment variable (recommended)

export PERPLEXITY_API_KEY="pplx-..."
vibecli --provider perplexity

Option 2: Config file (~/.vibecli/config.toml)

[perplexity]
enabled = true
api_key = "pplx-..."
model = "sonar-pro"

Model Selection

Model Strengths Best for
sonar-pro Deep research, citations Complex research tasks
sonar Fast search-augmented Quick lookups

Default: sonar-pro

Override from the CLI:

vibecli --provider perplexity --model sonar

Best For

  • Research-heavy tasks – find current documentation, API references, library versions
  • Debugging with context – search for error messages and known issues in real-time
  • Technology comparison – evaluate libraries and tools with up-to-date information
  • Learning new APIs – get current examples and best practices

Verify Connection

vibecli --provider perplexity -c "What are the latest changes in React 19?"

Troubleshooting

Invalid API key

Slow responses

  • Search-augmented models perform web searches before responding
  • Expect 3-10 seconds for research-heavy queries