xAI launches Grok 3 AI, claiming it is capable of 'human reasoning'

xAI has launched its Grok 3 models, which are reportedly an order of magnitude more capable than the previous Grok 2 models. The Grok 3 mini model can answer questions quickly but lacks accuracy, while the Grok 3 Reasoning and Grok 3 mini Reasoning models can mimic human-like reasoning. These models are capable of analyzing information and providing accurate answers. Other AI models with similar capabilities include DeepSeek's R1 and OpenAI's o3-mini. xAI claims that Grok 3 Reasoning outperforms o3-mini on several benchmarks. The Grok 3 features will be available to subscribers of X's Premium+ tier, which now costs $40 a month in the US. A separate subscription option for the standalone Grok app and Grok on the web will also be available, reportedly called SuperGrok and costing $30 a month. With Grok 3, users can ask the chatbot to "Think" to tap its reasoning capabilities for complex questions, and use the "Big Brain" function for even more complex queries. The models' reasoning capabilities also power a new feature called DeepSearch, which scans the internet and X to provide a brief summary for research inquiries. Additionally, xAI announced that the Grok app will get a "voice mode" within a week, and Grok 2 will be open sourced in the coming months.