OpenAI: o3 Mini

OpenAI o3-mini, released on January 31, 2025, is a cost-efficient language model specifically optimized for STEM reasoning tasks. It demonstrates moderate speed performance, ranking in the 38th percentile across benchmarks, and offers moderate pricing, placing it in the 22nd percentile. A standout feature is its exceptional reliability, achieving the 100th percentile with minimal technical failures. The model excels in specialized areas, achieving 93.0% accuracy in Coding and 98.0% in Reasoning, placing it in the 91st and 95th percentiles respectively. This aligns with its design for STEM tasks, where it matches the performance of the larger o1 model on challenging evaluations like AIME and GPQA with medium reasoning effort. General Knowledge and Ethics also show strong accuracy at 99.5% and 99.0%. While its Instruction Following accuracy is moderate at 59.6%, its Email Classification is high at 98.0%. The model's ability to adjust reasoning effort ("high", "medium", "low") provides flexibility for balancing performance and cost. Key developer capabilities include function calling, structured outputs, and streaming, though it lacks vision processing. Expert testers preferred o3-mini's responses 56% of the time over its predecessor, noting a 39% reduction in major errors on complex questions.