AI Model Reviews
LLM benchmarks are terrible. Everyone overfits their models so they can max out benchmarks in no more than a few months after its release. Open source models release with headlines "90% of Opus at 5% of the cost", yet anyone who has actually …