Each LLM provider has unique quirks and issues. This section covers common provider-specific issues and how to resolve them.

​
OpenAI

  1. Organization IDs: Set the organization ID if using a shared account
// @doctest id="fe5c"
// In config/llm.php
'metadata' => [
    'organization' => 'org-your-organization-id',
],
  1. API Versions: Pay attention to API version changes
// @doctest id="276a"
// Updates to OpenAI API may require changes to your code
// Monitor OpenAI's release notes for changes

​
Anthropic

  1. Message Format: Anthropic uses a different message format
// @doctest id="58e7"
// Polyglot handles this automatically, but be aware when debugging
  1. Tool Support: Tool support has specific requirements
// @doctest id="fe91"
// When using tools with Anthropic, check their latest documentation
// for supported features and limitations

​
Mistral

  1. Rate Limits: Mistral has strict rate limits on free tier
// @doctest id="dcd0"
// Implement more aggressive rate limiting for Mistral

​
Ollama

  1. Local Setup: Ensure Ollama is properly installed and running
# @doctest id="ee4f"
# Check if Ollama is running
curl http://localhost:11434/api/version
  1. Model Availability: Download models before using them
# @doctest id="4d47"
# Pull a model before using it
ollama pull llama2