Puck.Backends.ReqLLM (Puck v0.2.11)

Copy Markdown View Source

Backend implementation using the ReqLLM library.

ReqLLM provides a unified interface to multiple LLM providers.

Model Specification

The model must include the provider in "provider:model" format:

model: "anthropic:claude-sonnet-4-5"

Configuration

Backend options are passed in the client tuple and forwarded to ReqLLM:

  • :temperature - Sampling temperature (0.0 to 2.0)
  • :max_tokens - Maximum tokens in response
  • :top_p - Nucleus sampling parameter
  • :stop - Stop sequences

Examples

# Basic usage
client = Puck.Client.new({Puck.Backends.ReqLLM, "anthropic:claude-sonnet-4-5"})

# With options
client = Puck.Client.new(
  {Puck.Backends.ReqLLM, model: "anthropic:claude-sonnet-4-5", temperature: 0.7}
)

See ReqLLM documentation for supported providers and additional options.