Moonshot paused new Kimi K3 subscriptions after demand pushed GPU capacity to its limit, protecting existing users while adding compute. The launch of the 2.8T-parameter model hit US tech stocks.
Chinese AI startup Moonshot paused new subscriptions to its Kimi K3 open-weight model after demand pushed GPU capacity to the limit, the company said Sunday.
"Kimi K3 has received far more love than we expected, and our GPUs are feeling it," Moonshot said in a post on X. Over the previous 48 hours, demand had pushed close to the limits of current capacity. The company said it was adding compute as fast as possible and would reopen subscription slots in batches. Existing subscribers are not affected.
Moonshot plans to split membership into two tiers: Kimi Membership for web and app use and Kimi Code Membership for coding workflows. "This will help us match compute more precisely and keep the experience stable," the company said.
The capacity crunch follows the July 16 launch of Kimi K3, a 2.8 trillion-parameter open-weight model with native vision and a 1 million-token context window. Moonshot described it as the first open 3T-class model, designed for long-horizon coding and reasoning. Moonshot said Kimi K3's overall performance trails proprietary systems such as Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol. Internal benchmarks showed frontier-level results, the company said.
The model is the world's largest open-source AI model by parameter count, the Associated Press reported. Its release rattled US AI stocks. Shares of American tech giants slipped on the news, the AP said, because a cheaper, capable open-weight model from China could undercut their pricing power and demand. Reuters said the launch challenged assumptions that Chinese model developers trail their US peers by months.
Kimi K3 is an open-weight model, meaning other companies and developers can run it on their own hardware. Moonshot's own cloud service, which hosts the model for subscribers, hit capacity limits. The company is adding NVIDIA GPUs to expand, it said.
The new membership tier structure aims to allocate compute resources more efficiently, directing GPU cycles to coding workloads that may require more intensive processing. The Kimi Code Membership targets developers working on coding tasks, while the general membership covers lighter use cases.
The pause on new Kimi K3 subscriptions remains in effect until capacity expands. Moonshot said it would add capacity in batches and notify users when spots open.
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.