Privacy Policy
Effective October 12, 2026
Tryingopen is run by Offline Base. This policy covers tryingopen.com, including Self Chat at tryingopen.com/self-chat, the provider sites that run on the same service (tryingcoreweave.com, tryingbaseten.com and tryingmodal.com), and the Tryingopen browser extension. “We” and “us” mean Offline Base.
The short version
- The chat has no accounts. We never ask for your name or email, unless you sign up for the optional developer API.
- We do not store your conversations on our servers.
- Your messages go to the AI provider that serves the model you pick, with a zero-data-retention requirement wherever one is available. The model picker labels the exceptions.
- In Self Chat, your chats are saved only in the place you choose: your Google Drive, OneDrive or Dropbox, or a folder on your own computer.
- We do not sell your information, use it for advertising, or use it to train AI models.
What happens to your messages
When you send a message, your browser sends it to our server, which passes it on to be answered and streams the reply back. We do not log the text of your messages or of the replies.
- Many models are reached through OpenRouter, which sends each message to a company that runs the model. We attach a zero-data-retention requirement, so only a provider that promises to delete the message after answering, and not to train on it, may serve it. If none will, the message fails rather than being sent anyway.
- Other models are reached through an AI gateway run by ngrok instead, unless you pick a provider for them. Whether those keep data depends on the company serving the model, and any that may keep it are labelled “Provider may log” in the model picker. Picking a provider in the composer sends the message through OpenRouter, as above, with the zero-data-retention requirement.
- Liquid AI LFM3-VL 2B, the default pick on tryingopen.com, and the other edge models (Nemotron 3.5 Bolt and Liquid AI LFM3 1.7B) run on Offline Base’s own devices, reached through an ngrok tunnel. A small routing model from Liquid AI, which we run ourselves on one of those devices, reads each message to decide whether our device answers it or a cloud model you choose does (through OpenRouter, as above). That decision is made on our own hardware; no other company is sent your message to make it. Our devices record the size and timing of requests, never their text. When you pick a cloud model instead, your first two questions in a chat are also answered by our devices, so you can compare the two.
- Web search. When a reply needs the web, the search query goes to a search provider through OpenRouter. This sits outside the zero-retention requirement. In Self Chat, search is off until you switch it on; with it on and a model on your own computer, your latest message is sent to our server to be looked up, and only the results go back to your computer.
- PDFs. A PDF you attach is read by OpenRouter’s file parser, which also sits outside the zero-retention requirement. Self Chat does not accept PDFs.
- Compare with a closed model. Only if you press it, the conversation up to that reply is sent to the closed model’s maker (Anthropic or OpenAI) through ngrok’s gateway, under that company’s own terms.
- The browser extension sends the text you highlight, and the conversation it came from, in the same way.
What we keep
- Your IP address, briefly. We count how many messages each IP address sends per hour, to keep the free service available to everyone. The counts are held by Upstash and expire on their own within about two hours.
- Page views. We use Vercel Web Analytics, which counts visits without cookies and without identifying you.
- Hosting logs. Our host, Vercel, keeps standard request logs (such as IP address, time and the page requested) for a limited period. They do not contain your messages.
Nothing else about you is stored on our servers, unless you sign up for the developer API (below).
What stays in your browser
- Preferences such as your theme, your picks in the composer, and whether you closed a notice, kept in your browser’s local storage. The chat sets no cookies; the developer API sets one sign-in cookie, and only if you log in to it.
- Share links carry the conversation inside the link itself, after the “#”, a part of a web address that browsers never send to any server. There is no copy with us, and anyone you give the link to can read it.
- Files the model creates, their previews, and exports are put together in your browser and saved straight to your device.
The developer API (optional)
At tryingopen.com/developers you can sign up to use these models from your own code. Only if you do, we keep:
- Your email and a hash of your password (never the password itself). We do not verify the email or write to it.
- A sign-in cookie, valid for 30 days or until you log out.
- The names you claim, and which models each may use.
- Usage records: for each name’s last 20 requests, the time, model, provider, status, token counts and credits used, kept for up to 90 days so you can see them on your dashboard; and each day’s request and credit counts per name and per model, and your account’s running total of credits used (which enforces the daily credits), kept for two days. Not the messages or the replies, which pass through as they do in the chat and are not stored by us. As in the chat, a model marked “Provider may log” on the developers page goes to a provider that may keep them.
These are held by Upstash. Deleting a name deletes its usage records. To delete your account, write to us.
Self Chat: chats kept in your own storage
Self Chat saves each chat in a folder in the place you connect, and nowhere else. Your browser talks to that service directly: your files and your access token never pass through our servers. The access token is kept only for the browser tab you are using.
- Google Drive. We ask only for the
drive.filepermission, which lets Self Chat see and change only the files it created. We use it solely to create, read, update and move to the trash your chat files in the “Open Chats” folder, so you can keep and reopen your chats. - OneDrive and Dropbox, where offered, give Self Chat one app folder of its own and nothing else in your account.
- A folder on your computer. Self Chat can read and write only the folder you pick. Nothing is uploaded.
Tryingopen’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. We do not use data from Google Drive, OneDrive or Dropbox for advertising, do not sell it, do not use it to train AI models, and do not let anyone read it.
Files kept in Google Drive, OneDrive or Dropbox are held by that company under its own privacy policy. Disconnecting in Self Chat stops access from your browser; to remove Self Chat’s permission entirely, use your Google account’s third-party connections, your Microsoft account’s app permissions or your Dropbox connected apps. Deleting a chat’s folder deletes the chat.
A model on your own computer
If you connect a model running on your computer (for example in Ollama, LM Studio, llama.cpp or Jan), your browser sends your messages straight to it. They never reach our servers or the internet. Your browser remembers the server’s local address and its model names so they are there next time.
Who we share information with
Only the services named above, and only to do what you asked: answer a message, search the web, or save a chat where you chose. We may also disclose information if the law requires it. We do not sell or rent personal information.
Children
Tryingopen is not directed at children under 13, and we do not knowingly collect information from them.
Your choices
Because the chat has no accounts and we do not keep your conversations, we usually hold nothing that can be tied to you (the developer API, if you sign up, is the exception above). You can clear what your browser keeps by clearing this site’s data in your browser settings. For any question or request, write to us and we will help.
Changes
If we change what happens to your information, we will update this page and the date at the top.
Contact
Offline Base, team@offlinebase.com, offlinebase.com.