Moonshot AI

Moonshot AI (月之暗面) is a Beijing-based frontier AI lab founded in March 2023 by Yang Zhilin, previously a co-founder of Zhipu AI. It became China's fastest decacorn and has raised through a rapid succession of rounds backed by IDG Capital, Alibaba, Tencent, Meituan's investment arm, Tsinghua Capital and China Mobile.

Its product line runs on the Kimi brand: the Kimi chatbot on web and mobile, Kimi Code, a coding agent with terminal, VS Code, Cursor and Zed integrations plus MCP support, and Kimi Work, a desktop agent for macOS and Windows that runs the lab's agent-swarm system on local hardware. The chatbot first gained prominence on long-context document analysis, and long context has remained a defining thread through every generation since.

The company's strategic identity is open weights at frontier scale. Where most labs treat open release as a tier below their flagship, Moonshot publishes its frontier models and prices API access aggressively enough to exert real cost pressure on US vendors. It is also one of the few Chinese frontier labs to disclose consumption figures publicly, and its models rank consistently among the most-used on OpenRouter. Its current flagship is a very large mixture-of-experts model that activates a small fraction of its experts per token, with a million-token context and native vision, and it is the largest open-weights model any Chinese lab has produced — beating the leading US models on much of the published coding and agentic suite while the lab concedes it still trails them overall, naming user experience rather than raw capability as the remaining gap.

The licensing shift that came with it matters beyond the legal text. Earlier generations shipped under Modified MIT terms; the current flagship's weights are public but under a custom license requiring a separate agreement for inference or fine-tuning services above a revenue threshold, and mandating Kimi branding in any product above a user or revenue ceiling, with internal use and Moonshot's own products exempt. Pricing moved with it: the lab is no longer undercutting the frontier the way earlier Chinese open models did, and is pricing like a frontier lab with the aggressive rate reserved for cached input. The lab whose strategic identity is open weights at frontier scale attached commercial strings at exactly the point its model became competitive with the proprietary frontier — and it did so as open source became explicit Chinese national strategy, which makes Moonshot the most concrete evidence available for a policy position China's leadership now articulates directly. Its next release tested that identity from the safety side: Kimi K3.1 Pro, a heavily post-trained K3 launched in October 2026, had cyber-related data stripped from its training before its promised open-weight release, under a stated rule that Moonshot will not publish open weights that extend the frontier of offensive cyber capability.

The company also sits at the center of the distillation accusations against Chinese labs. A September 2026 joint advisory from the US National Security Agency, Federal Bureau of Investigation and Cybersecurity and Infrastructure Security Agency named Moonshot among six Chinese companies it said had extracted outputs from US frontier models at industrial scale. Anthropic's own telemetry, published two days later, credited Moonshot with more than 23 million exchanges with Claude and alleged it had relayed its customers' requests to Claude without telling them. On September 30, 2026 OpenAI attributed a core cluster of a separate campaign to extract its models' hidden reasoning to individuals associated with Moonshot. Both vendors are competitors, and Moonshot has not publicly answered the claims, but two labs' independent logs now name the same company.