Terms of Service

The service

OpenPeach is a personal-assistant platform you are not charged for. It orchestrates a Large Language Model that you supply (a cloud API key, or a model running on your own machine). You are responsible for your own provider costs — OpenPeach never bills you, and any usage figures we show mirror what your provider will charge.

Your responsibilities

• Use the service lawfully and don’t abuse it (e.g. spam, automated scraping, or attempts to disrupt it). We apply rate limits and may suspend accounts that abuse the platform.
• You’re responsible for the content you store and the instructions you give Peach, including anything you have it do on a connected machine (peachd) or via connected tools (MCP servers).
• Comply with the terms of whatever model provider you connect.

Model output

Large Language Models can be wrong. Peach’s output is assistance, not professional advice — verify anything important. You’re responsible for actions you or Peach take based on it.

Training our models — optional, and off unless you ask

Accepting these Terms does not turn training on. Contributing your conversations is optional, it has its own box on the signup form, that box starts unticked, and you can create and use an account without ever ticking it — nothing about the service changes either way. If you do turn it on, at signup or later in Settings → “Help train OpenPeach’s models”, it means we may use your Peach Chat and Ask Muse conversations and your Vibe Peach coding turns — the conversation, including the reasoning and tool calls a coding turn makes in the project you pointed it at, and the code it read and wrote there — to train our own Large Language Models. You can switch it back off at any time in the same place.

We never train on your files. The documents and files you upload, their contents, your Bio Vault, your health data and your document scans are never training data; a quote, citation or summary of one of your files that appears in a reply is part of the conversation and is eligible. We never sell your data or use it for any purpose other than training our own models. Contributed turns are kept 5 years while training is on and 30 days after you switch it off, and you can delete everything you have contributed at any time, separately from the switch. Personal details are removed by a pattern-based scrubber before a turn is used for training — at the curation step, not at capture — and that scrubber removes emails, phone numbers, card numbers, IP addresses and keys, but not names or postal addresses. The full, precise description, along with the lawful basis and your rights, is in the Privacy Policy.

Availability & changes

The service is provided “as is” during this early period and may change or have downtime. You can export your data or delete your account at any time from Settings. We’ll post material changes to these terms here.