Just days after unveiling what it calls its most advanced artificial intelligence model yet, Chinese AI start-up Moonshot has been forced to temporarily stop accepting new paid subscribers. The reason? Kimi K3 has become too popular. According to the company the demand for its latest AI model has overwhelmed the company’s computing infrastructure.
In a statement on X, the company wrote “Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we’re temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected.”
The company adds that existing subscribers will not be affected and that new memberships will reopen in batches as more computing capacity comes online.
AI subscription model also changes
Alongside the temporary pause, Moonshot has also announced changes to its subscription model. “Going forward, we’ll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable,” it wrote.
The subscription pause comes just days after Moonshot introduced Kimi K3 at the World Artificial Intelligence Conference in Shanghai. The company is positioning this new model as one of China’s strongest challengers to leading US AI models from OpenAI and Anthropic. The company also plans to release K3’s open-source model weights on 27 July, allowing developers to download, customise and run the model themselves.
What is Kimi K3?
Moonshot describes the Kimi K3 as its most capable flagship model to date. Built on a 2.8-trillion-parameter Mixture-of-Experts (MoE) architecture, the model features a one-million-token context window, native multimodal capabilities for understanding both text and images, and is designed for long-running reasoning, coding and AI agent tasks.
Unlike traditional chatbots that mainly answer questions, Kimi K3 is built to complete longer and more complex workflows with minimal human supervision. Think of it as an AI assistant that can carry out multi-step tasks instead of simply responding to one prompt at a time. Moonshot says the model can help write software, analyse research papers, work through financial data, understand images, create presentations and even assist with video editing.
Powering all of this are two new technologies that Moonshot calls Kimi Delta Attention and Attention Residuals. The company says these technologies allow the model to process much larger amounts of information while using computing resources more efficiently.
Competitor to Claude and ChatGPT
Moonshot is positioning Kimi K3 as a serious competitor to AI models from OpenAI and Anthropic. The company says K3 performs at or near the level of leading frontier AI systems on several public benchmarks. It also claims that Kimi K3 Max outperformed Anthropic’s Claude Opus 4.8 and OpenAI’s GPT-5.5 in some of its own internal tests covering online research, presentation creation and financial analysis. However, it is worth noting that these results come from Moonshot’s own evaluations and have not yet been independently verified.
The launch comes as Chinese and US companies race to build more powerful AI models. Unlike OpenAI and Anthropic, which keep their flagship models closed, Moonshot plans to make Kimi K3 open-source so developers can download, customise and build on it.
