OpenAI employees accused xAI's released Grok3 AI model Benchmark test results of being misleading, but xAI denied any wrongdoing. Grok3 outperformed OpenAI's strongest model at AIME 2025, but it was alleged that the chart did not include all comparison data. Babushkin argued that OpenAI had also released similar charts in the past for comparing model performance.

Recently, OpenAI employees accused xAI of misleading with the Grok3Benchmark AI model test results, claiming outstanding performance but omitting key data. The two sides engaged in heated discussions on social media, with xAI claiming that OpenAI had previously released similar charts.

On February 18, Jinshi Data learned that while Musk released three major Grok models, the Grok website was inaccessible.