Microsoft unveils two new large language models: MAI-Thinking-1 and MAI-Code-1-Flash. MAI-Thinking-1 boasts 1 trillion parameters with 35 billion active, showing strong reasoning capabilities and outperforming Claude 3 Sonnet in human evaluations. MAI-Code-1-Flash, a 137 billion parameter model with 5 billion active, is optimized for code generation tasks, enhancing GitHub Copilot performance and cost-efficiency.
Opening Kapyn…