Language models fine-tuned for advanced reasoning tasks. Optimized from pretrained architectures for complex problem-solving.