Aleph Alpha's Kolibri-1 playing Breakout – Making decisions under 25ms per move (twitter.com)

Aleph Alpha has successfully demonstrated its AI model, Kolibri-1, playing the classic arcade game Breakout independently, with no fine-tuning required. This achievement highlights the model's capacity for real-time decision-making, functioning at an impressive inference speed of around 25 milliseconds per move. The innovative approach involves optimizing action probabilities, allowing Kolibri-1 to execute four distinct moves without generating text, thus emphasizing its practical application in game-based scenarios. This milestone is significant for the AI/ML community as it opens new avenues for research and development in autonomous systems and real-time AI applications. By providing open weights, Aleph Alpha fosters collaboration and experimentation among developers, deepening the dialogue around sovereign AI. The launch coincided with German Unity Day, underscoring a sense of national pride in contributing to the future of AI. The model is now available for free at tesseracted.com, inviting users to explore its capabilities further and sparking excitement about its potential in diverse applications.