Anthropic Claude

Anthropic Claude Provider Options

Weaver supports provider-specific configuration options for Anthropic Claude models. These can be set in the Provider Options (JSON) field in the chat configuration settings.

Available Provider Options

Thinking Mode (Reasoning)

Enable Claude's reasoning capabilities for more thoughtful responses.

{
  "thinking": {
    "type": "enabled",
    "budgetTokens": 12000
  }
}

Parameters:

type: Set to "enabled" to activate thinking mode
budgetTokens: Maximum number of tokens Claude can use for reasoning (recommended: 12000)

This feature is currently available for claude-3-7-sonnet and provides access to Claude's step-by-step reasoning process before generating responses.

When enabled, Claude will perform extensive thinking about complex questions before answering, resulting in more accurate and well-considered responses.

Benefits of Thinking Mode

More accurate factual responses
Better structured problem-solving
Enhanced reasoning for complex questions
More nuanced exploration of topics

Model Compatibility

Check the Anthropic documentation for the most up-to-date compatibility information. AI SDK

Anthropic Claude