LLM360/K2-Think

K2-Think is a 32 billion parameter open-weights general reasoning model developed by Zhoujun Cheng et al. It is specifically designed for strong performance in competitive mathematical problem solving and general reasoning tasks. The model features a large 131072-token context length and demonstrates high inference speeds on specialized hardware. It excels in complex math benchmarks like AIME and HMMT, making it suitable for advanced analytical applications.