The xAI Grok-3 Mini model successfully passed a difficult test on logical thinking. In the Marcus Problem task, it scored 120 out of 120 points, interpreting tangled and mixed sentences perfectly. In the Alice+ Problem test, Grok-3 Mini also showed results with 24 out of 24 points, despite the presence of distracting data. In high-level mixed tasks, the model scored 24 out of 24 points, surpassing GPT-4.5 and Gemini 2.5 Pro models. Elon Musk noted: 'Soon AI will far exceed the best people in reasoning.'