Replicate

replicate

YAML config

providers:
  - id: replicate:<your-model>
    config:
      apiKey: ${REPLICATE_API_KEY}

TypeScript usage

import { createProvider } from "@evalguard/core";

const provider = createProvider("replicate", process.env.REPLICATE_API_KEY);
const response = await provider.complete({
  model: "<your-model>",
  messages: [{ role: "user", content: "Hello" }],
});

Authentication

Set REPLICATE_API_KEY in your environment. EvalGuard validates the key on first call and surfaces typed errors for 401 / 403 / rate-limit responses (with Retry-After parsing).

Setup walkthrough

  1. 1. Sign up for a Replicate account at replicate.com.
  2. 2. Obtain your API key from the account settings.
  3. 3. Install the Replicate Python client using 'pip install replicate'.
  4. 4. Authenticate your API key in your application using 'replicate.Client(api_token='YOUR_API_KEY')'.
  5. 5. Choose a model from the Replicate model list and note its model ID for use.

Gotchas

  • Rate limits are enforced; ensure you monitor your usage to avoid hitting limits.
  • Some models may have regional restrictions, so verify availability in your area.
  • API responses can vary in latency; implement retries for better reliability.

Cost note

Pricing for the cheapest models starts at $0.01 per 1,000 tokens, while flagship models can go up to $0.20 per 1,000 tokens.

Recommended models

Eval / judge
gpt-3.5-turbo
Agent / tool-use
gpt-4
Code
code-davinci-002
Vision
stable-diffusion

LLM-generated (llm-gpt-4o) · 2026-05-23