Moonshot AI pauses Kimi K3 subscriptions as demand overwhelms compute

Moonshot AI, the Chinese AI lab behind the Kimi line of large language models, has temporarily halted new subscriptions for its Kimi K3 model. The decision, reported on August 7, 2026, came after overwhelming user demand pushed the company's computing infrastructure to its breaking point. Kimi K3 is a massive open-weight model with 2.8 trillion parameters, making it one of the largest publicly accessible AI systems available. Its API launched in mid-July and quickly attracted a wave of developers and enterprises eager to run or fine-tune the model on their own infrastructure.

This pause underscores a growing challenge for AI vendors: even with cutting-edge hardware, serving a 2.8T-parameter model at scale is extraordinarily compute-intensive. Moonshot has stated that existing subscribers will continue to have access, while new signups are temporarily blocked until capacity is expanded. The incident highlights the enormous appetite for open-weight models, which offer transparency and customization that closed APIs cannot match. For Moonshot, the demand is a validation of its technology but also a logistical test — one that may prompt accelerated infrastructure investment and smarter resource allocation in the coming months.