New Conversation
Approaching context limit — consider starting a new chat
FuddiG AI
Your intelligent multi-model chat interface
Your intelligent multi-model chat interface
Save multiple keys locally, name them, and choose which one is active with the radio selector.
Local proxy base URL. This app sends requests here so the browser never talks to the remote API directly.
Edit or reset today's spend for the active API key.
A short persona description injected into the system prompt (e.g., tone, role, specialties).
A template to wrap user prompts (use {{user_prompt}} as placeholder).
Include example exchanges before the user prompt to guide model behavior.
One or more helper examples (separate by `---`). These will be prepended when few-shot is enabled.
Choose the default tone for system instructions.
Limit how many tokens the system prompt may consume.
Default model for new conversations.
Fast, cheap model used to auto-generate conversation titles.
Maximum tokens in the model's response.
Controls randomness. Lower = focused, higher = creative.
Enable extended reasoning by default for new chats.
Max tokens reserved for internal reasoning steps.
Injected as the first system message in every conversation.
Tighter spacing between messages.
Press Enter to send. Shift+Enter for newline.
Display token count under each assistant message.
Automatically name new conversations after the first exchange.
Download all conversations as a JSON file.
Merge conversations from a previously exported JSON file.
All keyboard shortcuts available in FuddiG AI.
| CtrlN | New conversation |
| CtrlK | Command palette / search |
| Ctrl, | Open settings |
| Enter | Send message |
| ShiftEnter | New line in input |
| CtrlEnter | Send message (always) |
| CtrlShiftT | Toggle thinking mode |
| CtrlL | Clear current chat |
| CtrlShiftE | Export current chat |
| Escape | Stop generation / close modal |
| ↑ (empty input) | Edit last user message |
| ? or Ctrl/ | Show this shortcut list |
Are you sure?