Choose which Large Language Model (LLM) powers your content generation using the model parameter.

Each model offers different trade-offs between quality and speed. Currently, cost is the same regardless, though this may change in the future.

Model routing uses keyword matching — any value containing the keyword will route to that model. Models are updated to the latest stable versions regularly.

Model Options:

• gemini: Google's Gemini 2.5 Flash – fast, analytically strong, excellent structured output.

  • Best for: Analytical content, structured writing, deep context handling, large-scale generation
  • Keyword: Any model string containing gemini (e.g., gemini, gemini_flash, gemini_2.5)
  • Email generation: ~2-4 seconds

• sonnet: Anthropic's Claude Sonnet 4 – balanced quality and speed, strong instruction-following.

  • Best for:
    • General-purpose content
    • Multi-email sequences (conversational tone)
    • Complex formatting requirements
  • Keyword: Any model string containing sonnet (e.g., sonnet, sonnet_3.5, sonnet 3.5)
  • Email generation: ~3-5 seconds

• gpt4o: OpenAI's GPT-4o – reliable, fast, good for real-time applications.

  • Best for:
    • Fast, dynamic content
    • Website personalization
    • Default behavior when no model specified
  • Keyword: Any model string containing gpt4o (e.g., gpt4o, gpt4o-latest)
  • Email generation: ~2-5 seconds

📘

Default behavior: If no model parameter is provided (or an unrecognized value is passed), the system defaults to GPT-4o.

🔄

Fallback: If the selected model experiences an error, requests automatically retry and then fall back to Claude Sonnet 4 to ensure delivery.

How to use the model parameter

When making a content generation request, the model parameter can be explicitly set. The system matches on keywords — you don't need to specify exact model versions.

{
  "contactEmail": "[email protected]",
  "userEmail": "[email protected]",
  "contentType": "custom",
  "model": "gemini"
}

All of the following would route to Gemini:

"gemini"
"gemini_flash"
"gemini_flash_1.5"
"gemini-2.5-flash"

Breakdown by contentType

⚠️

Deprecated model values: opus, opus2, gemini_flash_1.5, gpt5 — these legacy values still work (they route to the default GPT-4o path) but are no longer recommended. Use the keyword-based values above.

Legacy Compatibility

For backward compatibility, the following exact legacy values continue to work:

Contact [email protected] to get started.