provider-replicate/quickstart (Replicate Quick Start)

You can run this example with:

npx promptfoo@latest init --example provider-replicate/quickstart
cd provider-replicate/quickstart

This is a minimal example to test the Replicate provider with promptfoo.

Quick Start

1. Get a Replicate API token from https://replicate.com/account/api-tokens

2. Set the environment variable:

   bashCopy

   export REPLICATE_API_TOKEN=r8_your_token_here
   

3. Run the evaluation:

   bashCopy

   promptfoo eval
   

What This Tests

This example uses Replicate's "hello-world" model - a simple, fast model that's perfect for testing your setup. It:

• Verifies your API key is working
• Tests basic prompt templating
• Demonstrates simple assertions
• Runs quickly (< 5 seconds per test)

Next Steps

Once this works, try:

1. Different Models: Replace the model ID with:

   • replicate:meta/meta-llama-3-8b-instruct - For text generation
   • replicate:image:stability-ai/sdxl:7762fd07cf82c948538e41f63f77d685e02b063e37e496e96eefd46c929f9bdc - For image generation (or use latest version from Replicate)

2. More Examples: Check out:

   • provider-replicate/llama4-scout - Advanced text generation with Llama 4
   • provider-replicate/comprehensive - All provider features
   • provider-replicate/image-generation - State-of-the-art image generation

Troubleshooting

• API Key Error: Make sure REPLICATE_API_TOKEN is set correctly
• Model Not Found: Check the model ID matches one from https://replicate.com/explore
• Timeout: Some models take 30-60 seconds on first run (cold start)