nvidia/OpenMath-Nemotron-1.5B

OpenMath-Nemotron-1.5B by NVIDIA is a 1.5 billion parameter, Qwen2.5-based transformer decoder-only language model, fine-tuned on the OpenMathReasoning dataset. It is specifically optimized for advanced mathematical reasoning tasks, achieving state-of-the-art results on benchmarks like AIME24, AIME25, and HMMT-24-25. This model supports a context length of up to 131,072 tokens and is designed for commercial use in mathematical problem-solving.