According to Foresight News, AI startups Flower AI and Vana have jointly introduced a distributed large language model called Collective-1. Unlike traditional models that rely on centralized data centers, Collective-1 runs on GPUs distributed across various global locations. It also integrates private user data from platforms such as X, Reddit, and Telegram. With 7 billion parameters, the model is smaller than top-tier systems like ChatGPT and Claude. However, Flower AI is planning to scale up, with a 30-billion-parameter model in development and a nearly 100-billion-parameter version expected later this year.

In partnership with universities in China and the UK, Flower AI has also developed Photon, a tool designed to improve the efficiency of distributed training. This strategy aims to make foundational AI development more accessible to smaller businesses and countries with limited resources. Meanwhile, Vana offers tools that give users control over how their data is used, including the ability to grant permissions and receive benefits. Experts suggest this collaborative and decentralized approach could signal a shift toward a more open future for AI.