Moonshot AI's consumer AI assistant — Kimi Web, the Kimi app, Kimi Work, and Kimi Code, now running on the K3 model — whose new-subscription signups were paused in mid-July 2026 after demand for K3 outran its GPUs.
Last verified · 2026-07-19 · by Moe Ameen
Kimi is the consumer-facing AI assistant from Moonshot AI, the Beijing lab behind the Kimi model series. It is the product you actually use rather than the model underneath it: a chat assistant on the web at kimi.com and in the Kimi mobile app, plus two work-oriented surfaces — Kimi Work, an agentic assistant for documents, research, and multi-step tasks, and Kimi Code, a coding-focused workflow. It now runs on Kimi K3, Moonshot's flagship frontier model, which brings a 1-million-token context window and native image understanding into the assistant.
In mid-July 2026, days after the K3 launch, Moonshot paused new Kimi subscriptions. In a public post it said K3 had drawn far more demand than expected — over roughly 48 hours, usage pushed close to the limits of its current capacity — so it was temporarily halting new signups to protect the experience of existing subscribers while it added GPU capacity, and it said it would reopen new-subscription spots in batches. Moonshot also said it would split membership into two more focused plans: a Kimi Membership covering Kimi Web, the app, and Work, and a separate Kimi Code Membership for coding workflows, so it can match compute to usage more precisely. Existing members kept access throughout; the pause applied to new paid signups.
As a product, Kimi sits in the same lane as the ChatGPT, Claude, and Gemini assistants: you prompt it and it reasons, drafts, reads documents and images, and works agentic tasks. Its long context is a real draw — you can feed a whole transcript, a research folder, or a codebase into a single conversation. The honest framing: Kimi is an assistant you talk to, not a content studio. It produces text, code, and analysis; it renders no finished social video, no branded image, no carousel, and it publishes to no platform. And as the mid-July pause showed, access to any single hosted assistant can be rationed by one lab's capacity at exactly the moment the tool gets popular.
The mid-July subscription pause is a useful lens on where Kimi fits. Kimi is where you think out loud: its 1M-token context lets you paste an hour-long transcript or a folder of research into one chat and pull back an outline, a batch of angles, or a script. What it hands you is words. It renders no captioned vertical short, no face-locked avatar video, no brand-exact carousel, and it posts nothing — and that is the whole second half of a creator's job. Kompozy is that half: drop a Kimi draft or brief in as a source and Kompozy generates finished media across 18 formats — Persona Shorts and HeyGen avatar video, Clipped Shorts, Carousels and Quote Graphics via brand-exact HyperFrames, Photo Posts, a Blog Article, and an Email Newsletter — every output held to one voice by your Persona Brief instead of reading like raw assistant output.
There is a second reason the pairing matters, and the signup freeze makes it concrete: when your workflow is chained to one hosted assistant, you inherit its capacity limits. New Kimi members were locked out for exactly as long as K3 was too popular to serve. Kompozy does not put you in that position — it runs its own managed Claude and OpenAI models for generation, so the engine keeps producing and scheduling regardless of whether any single lab is rationing GPUs. Use Kimi for its long-context reasoning on the days you can get in; let Kompozy turn the result into a scheduled, on-brand week across nine social platforms plus blog and email — and keep shipping even when one model is temporarily unavailable.
As of mid-July 2026 Moonshot temporarily paused new Kimi subscriptions after demand for its K3 model outran GPU capacity over about 48 hours. Existing subscribers were unaffected, and Moonshot said it would add capacity and reopen new-subscription spots in batches. Check kimi.com for current availability, since this is a moving situation.
Moonshot said the Kimi K3 launch drew far more demand than expected, pushing compute close to the limits of its current capacity within roughly 48 hours. To protect the experience of existing members it paused new signups while expanding GPU capacity, and it said it would reopen spots in batches and split membership into two plans — Kimi Membership (Web, app, Work) and Kimi Code Membership (coding).
Kimi is the assistant product — the web, app, Work, and Code surfaces you interact with. Kimi K3 is the underlying frontier model that now powers it. You subscribe to Kimi; K3 is the intelligence running behind the chat window. See the Kimi K3 page for details on the model itself.
No. Kimi drafts text, reasons over long inputs, and reads images, but it renders no captioned video, branded graphics, or scheduled posts, and it publishes to no platform. To turn a Kimi draft into finished, on-brand content across nine platforms plus blog and email, pair it with a content engine like Kompozy, which generates 18 formats and handles publishing.
Alongside the subscription pause, Moonshot said it would split Kimi membership into two more focused plans: Kimi Membership for Kimi Web, the app, and Work, and a separate Kimi Code Membership for coding workflows. The stated goal is to match compute to usage more precisely and keep the experience stable. Confirm current plan details on Moonshot's own pages.