Comparing local large language models for alt-text generation
Cloud-based LLMs excelled in generating alt-text accuracy, while local models performed reliably but with some detail omissions.
The blog author aims to improve alt-text for 9,000 images using tested AI models.