Flower AI and Vana Launch Distributed Training Large Model Collective-1

Foresight News reports that, according to Wired, AI startup Flower AI has jointly launched a large language model called Collective-1 with Vana, trained in a distributed manner.

This model runs on GPUs located worldwide, avoiding reliance on traditional data centers, and incorporates user-private data from X, Reddit, and Telegram.

Collective-1 currently has 7 billion parameters, significantly smaller than cutting-edge models like ChatGPT and Claude, but Flower states it is preparing to train a 30 billion parameter model and plans to release a version approaching 100 billion parameters within this year.