LLMWorkbook

LLMWORKBOOK Provider Anthropic

Anthropic’s Claude API has a distinct structure and accepts only specific parameters. This documentation explains how to configure and use the call_llm_anthropic function through the LLMConfig system in llmworkbook.

---

1. Anthropic Provider (call_llm_anthropic)

Configuration Keys in options:

• model

  • Type: str
  • Required
  • Description: Specifies the Claude model to use.

  • Example Values:

    • "claude-3-opus-20240229"
    • "claude-3-sonnet-20240229"
    • "claude-3-haiku-20240307"
  • Default: "claude-3-sonnet-20240229" (if not explicitly set)

• max_tokens

  • Type: int
  • Required
  • Description: Maximum number of tokens Claude can generate in the response.

• system_prompt

  • Type: str
  • Optional (set outside options)
  • Description: The system prompt guiding Claude’s behavior. Passed as system=... in API call.

---

✅ Example Configuration

from llmworkbook import LLMConfig

config = LLMConfig(
    provider="anthropic",
    system_prompt="You are a helpful assistant.",
    options={
        "model": "claude-3-haiku-20240307",
        "max_tokens": 512,
    }
)

---

API Key

• api_key

  • Your Anthropic API key can be:

    • Provided explicitly via config.api_key, or
    • Fetched automatically from the environment variable: ANTHROPIC_API_KEY.

---

This site is open source. Improve this page.