Moonshot AI has apologized to Kimi subscribers for rate limiting and unnecessary quota consumption after the launch of Kimi K3.1 Pro. The Beijing-based startup said it will reset every subscriber’s monthly quota by 25 percent.
Moonshot put the value of that reset in plain terms. It said the 25 percent reset equals one week of usage for subscribers. Subscribers who had used less than 25 percent of their quota will have their recorded usage reset to zero.
What changed for subscribers
The reset covers all subscribers, not only those who hit a limit. That matters because rate limits and extra quota use followed the launch for paying users. Every subscriber gets the same 25 percent adjustment.
Moonshot also adjusted parameters to cut waste. It reduced unnecessary subagent calls, where work is handed between AI agents, and trimmed output length, meaning fewer tokens generated per response. Both changes target the two sources of excess use that Moonshot named.
Why the launch strained quotas
Kimi K3.1 Pro is Moonshot’s most capable open-weight model to date, with a 1-million-token context window. It is available on Kimi.ai, the Kimi apps, Kimi Work, Kimi Code and the Kimi API. Its launch brought the rate limits and quota problems that the reset now addresses.
This is not the first time capacity has bitten Moonshot. The company previously faced computing capacity issues after the launch of Kimi K3. Reuters also notes that Chinese AI labs face computing shortages following United States chip export controls.
Moonshot still plans to release the full K3.1 Pro model weights by November 4. That open-weight release remains separate from the subscription fix. Subscribers get quota relief now, while developers waiting for the weights must wait for that date.
What subscribers should watch next
Subscribers can judge the fix by their own usage over the rest of the month. Steady quotas would show that the parameter changes worked. Repeated limits would show that capacity remains the binding constraint for the service.
For Moonshot, the episode shows how launch demand can outrun capacity, even for a model built for heavy agentic work. The quota reset repairs the immediate damage. The harder test is whether its infrastructure can match the models it ships.
