MerchantryTidbits

llm-inference / library

Replicate Python

Capability: Replicate Python

Use it when

  • run a published image, audio, or language model without hosting it
  • prototype against version-pinned community models

What it solves

Not the fit when

  • hosted model runs require an account and usage billing

Install

pip install replicate

Invoke

Set `REPLICATE_API_TOKEN`, then call `replicate.run(model_version, input={...})` or stream prediction events.

Alternatives

No reviewed alternatives recorded yet.