Is ArtificialAnalysis biased against open weight models?
It seems like whenever open weight models come close to proprietary US models, ArtificialAnalysis conveniently "re-calibrates" their benchmarks which then shoves open weight models down the toilet.