Why text-based evals fail for vision-language models

by nikhilpareek13 | Jan 6, 2026 | 0 comments on HN