Grok 3 mini Reasoning (high) is a compact language model developed by xAI, specifically optimized for logic-intensive tasks and STEM-based problem solving. Released in February 2025 as part of the Grok 3 family, it utilizes reinforcement learning to perform chain-of-thought reasoning. The "high" configuration refers to the model's highest reasoning effort level, which increases test-time compute to handle complex queries in mathematics, programming, and scientific analysis.
Despite its smaller size compared to the flagship Grok 3, the mini variant is designed for speed and cost-efficiency. It features a visible reasoning trace in its output, providing transparency into the model's internal thought process. This configuration is particularly effective for tasks requiring structured logic where a full-scale flagship model may be less efficient.
The model supports a context window of 131,072 tokens, enabling it to process large documents and complex codebases. It has demonstrated competitive performance on specialized benchmarks such as the American Invitational Mathematics Examination (AIME) and the Graduate-Level Professional Quality Assurance (GPQA) science dataset.