KelvinAI provides local-first AI coding and personal-assistant software, plus the hosted account service (model relay, credit billing, cross-device handoff, chat and collaboration).
You are responsible for activity under your account; do not sell or share it with untrusted parties.
No unlawful, infringing, fraudulent or malicious use; no reverse-engineering, attacking the service, or bypassing billing.
We may limit or terminate service for violations; unused credits follow the Refund Policy.
We have zero tolerance for unlawful or objectionable content — including sexual content, graphic violence, hate and discrimination, harassment and bullying, scams, illegal trade, and content that violates others' privacy or intellectual property — and for users who abuse or harass others.
In the app you can report inappropriate content or users (including AI-generated replies, images, video and audio), and block someone; once blocked, they can no longer send you messages.
We review and act on reports within 24 hours: violating content is removed, and accounts with serious or repeated violations are restricted or banned.
Public profile details such as nicknames, avatars and group names are checked automatically before they are shown; anything that fails the check must be changed before it can be used.
In line with China's Measures for Labeling AI-Generated Synthetic Content, we label AI-generated content: AI chat pages always show “You're chatting with AI — replies are AI-generated” (“You're chatting with AI” where space is tight), and pages that show earlier AI output, such as chat history, show “Content is AI-generated”; copying an AI reply, a screenshot answer or a screenshot translation appends “(This content is AI-generated)”; shared pages and share images say they were generated by AI; AI voice calls show that the voice is AI-synthesized; when AI voice input types a polished draft, its floating bar says the text was organized by AI; and when used in mainland China, generated images and videos carry an “AI生成” (AI-generated) mark in the bottom-right corner (except the desktop pet's appearance images, which Settings notes are generated by AI). In every region, generated image and video files also carry machine-readable information saying they are AI-generated; you cannot see it in the picture.
You must not remove, alter, forge or hide these labels, and you should state that content is AI-generated when you publish or distribute it.
Official credits are priced by actual model consumption (tokens), as shown on the recharge page.
Enterprise sub-accounts share the parent's credits; the parent may set monthly quotas.
We work to keep the service stable, but are not liable beyond fees paid for upstream-provider fluctuations or force majeure.
The software is provided as-is; local data (conversations, files) is your responsibility — back up what matters.