VibeFree privacy and AI providers

VibeFree sends every chat message to DeepInfra, a US company, which runs both models - DeepSeek V4.1 Flash and gpt-oss-20b - in the United States, and possibly its Canadian data centres, and does not store your prompt once it has answered, beyond a short-lived cache of the conversation (at most an hour) that only VibeFree's account can use. There is no router in between and no fallback host, so a message cannot end up on a host in China. If the optional Space Bunny Alpha stealth model is offered to you, it is the one exception: it is reached through OpenRouter, and its provider is not disclosed. Until 25 September 2026, DeepSeek V4.1 Flash ran on DeepSeek's own servers in the People's Republic of China. Two further providers are reached only when the assistant uses a specific tool: Google's Gemini API when you ask about an image, audio file or video, and TinyFish when it searches the web or reads a page. Cloudflare Workers AI, on VibeFree's own Cloudflare account, runs the ad classifier on each message you send, image generation and the search behind assistant memory.

  • DeepInfra keeps no copy of your prompts or the AI's responses beyond that cache; the Space Bunny Alpha provider may. VibeFree stores web and mobile conversations so they are there on your other devices; the VS Code extension keeps its history on your machine instead.
  • Only the content of your chat messages is sent to these providers - never your account details, credentials, or payment data.
  • Media you attach (images, audio, video) is analysed by Google's Gemini API, which holds it only for the duration of that analysis. VibeFree itself stores the file with the conversation until you delete it.
  • If you consent to external ad networks, your conversation is also sent to them - see how VibeFree stays ad-funded for who receives what.
  • Image Studio, in limited testing, makes images on Cloudflare Workers AI and DeepInfra, except its Pro model, which runs on BytePlus, part of ByteDance, in Singapore - and which it asks about before your first Pro image. See section 1.5h of the Privacy Policy.
  • When the assistant searches the web or reads a page, TinyFish performs that search or fetch, in the United States. It receives the query or web address the model wrote - which is drawn from your conversation - and never your IP address, because the request is made by VibeFree’s servers rather than your device.

Frequently asked questions

Which country processes my chat if I use DeepSeek V4.1 Flash?

The United States, and possibly Canada. Since 25 September 2026 it runs on DeepInfra, in the United States and possibly its Canadian data centres, and DeepInfra does not store your prompts beyond a short-lived cache of the conversation so far, kept for at most an hour. DeepSeek, the company that created the model, receives nothing. Before that date it ran on DeepSeek's own servers in the People's Republic of China.

Is any model processed in China?

No. Both models - DeepSeek V4.1 Flash and gpt-oss-20b - are run by DeepInfra in the United States (and possibly Canada), called directly with no other host in between. That holds in the web chat, the VS Code extension and the apps alike.

Does the assistant send my questions to a search engine?

Only when it decides it needs current information to answer you - a library version, a release note, an unfamiliar error. The search is run by TinyFish, in the United States, and it receives the query the assistant wrote plus nothing else. Because VibeFree's servers make the request rather than your device, TinyFish never sees your IP address. The query is written out of your conversation, so it can repeat details from your message - another reason not to put personal data in prompts.

Where can I read the full policy?

The complete Privacy Policy covers every data processor VibeFree uses, including Cloudflare, Stripe, Resend, Google and TinyFish, and the external ad networks that receive your conversation if you consent to them.

Related pages