We have been quietly building a new type of general frontier model since mid 2024. We are reinventing the entire LLM stack for real work, focused on reliability and decision-making. We are well funded with a small and fast moving team previously from OpenAI, Google Brain, Meta and Scale.