Meta's Maverick AI recently secured second place on LM Arena, a platform for evaluating AI model performance. However, controversy arose as researchers discovered that the tested version was an 'experimental' variant, not the public release. This has led to concerns about transparency and fairness in AI benchmarking. Critics argue that using a fine-tuned version for evaluation creates unrealistic expectations for developers. As calls for clarity grow, both Meta and the LM Arena have yet to respond to the backlash. The situation raises questions about the integrity of AI performance metrics.