REDPAPERS

Conversations

Your personal conversations appear here. Guest chats are not shown in this sidebar.

ADMINISTRATION

REDPAPERS

Model

Checking AI Model…

Please wait.

How can I help?

Ask a question or start a conversation.
Your messages remain in this conversation.

Thinking Web Max tokens

Enter to send · Shift+Enter for a new line

REDPAPERS

Site locked

Only an administrator can log in while the site is locked.

Log in

Need an account or access? Contact the administrator.

Guest password

Username Password

Admin settings

Model availability errors are shown in the conversation after you send a message.

AI model control

Checking model state...

Cancel loading waits for the model server to finish loading, then requests unload. It cannot force-stop the native loader.

Before Model ON, the server must verify available RAM on ai-ubuntu. If the model server does not expose RAM telemetry or free RAM is below the configured safety floor, loading is blocked. VRAM and final model fit cannot be guaranteed.

Lock site Allow guest login (auto-save)

Guest login is disabled.

Maximum output tokens

AI defaults — personal accounts

Thinking Default tokens Web mode Thinking budget (0 = unlimited, backend-dependent)

Allow personal users to change AI settings

AI defaults — guest

Thinking Default tokens Web mode Thinking budget (0 = unlimited, backend-dependent)

Allow guests to change AI settings

Thinking budget takes effect only when supported by the selected model.

Active inference jobs

Force Stop cancels REDPAPERS tasks, not the model service. Cancelling is not confirmation of GPU release.

Add user

Users

Guest password

Activity monitor

Online indicates recent activity. IP addresses do not establish identity.

Show

Select a session

Recent unsuccessful logins

Visitor analytics

Anonymous browser visits, submitted pre-login questions and approximate conversion. Bot/human labels are heuristics, not verified identities.

Select a visitor

Usage remaining

Questions and cumulative inference time are shared across requests. Either limit can stop new requests; time expiration stops an active job. These are server-side values.

Question & Time Management

Only Thinking level is configurable by guests and personal users. Total inference time is cumulative across questions; both limits apply. Stop active jobs before changing limits or resetting usage.