Mistral Medium 3.5 vs OpenAI: Are AI Benchmarks Rigged?

A recent YouTube Short by Zohaib Ehtisham questions whether the benchmark war between Mistral's Medium 3.5 model and OpenAI's offerings is rigged. The video, which garnered attention on Hacker News, notes that the performance gap is shrinking but asks whether local models can truly compete. As AI models become more capable, the validity of standardized benchmarks is increasingly debated, with concerns that some tests may favor certain architectures or training methods.