Examine the current model selection logic in the codebase, likely in a file handling OpenAI API calls. Add 'text davinci 003' and 'text davinci 002' to the list of supported models. Ensure that any rate limit handling or pricing considerations are accounted for. Test with a sample translation to verify functionality.