Hello, OpenAI team and community! 
I often use ChatGPT for one-off tasks: translation, string generation, text correction, brief explanations.
In such cases, I don’t need dialogue, context accumulation, or history preservation — just a single isolated response.
Currently, ChatGPT defaults to a dialogue using previous messages. Even when manually attempting to “reset” the chat through instructions, the model’s behavior remains unpredictable.
Suggestion
Add a stateless mode and incognito mode when creating a new chat, which does not use the current or previous history.
How it might look:
- When you click “+ New Chat” → additional options are displayed:
Stateless Chat — the model receives only the current message;
Incognito — the chat does not use previous history (including the “previous” incognito chat);
Hint: “Each message in this chat will be treated as a separate request.”
Convenience for technical users who don’t need dialogue — only results.
Possible implementations
-
Button:
+ New chat (StatelessCHAT) -
Item in the tool window (search the web, deep search, conduct research, etc.):
⛔️ Do not use history -
API parameter:
{
“context”: false,
“mode”: “stateless”
}
Summary
Stateless + Incognito mode is a fast, transparent, and honest way to interact with the model for one-step tasks.
This mode is ideal for many real-world use cases and will help reduce the load on the model.
Thank you for your attention! 
I would be delighted if this idea were taken into account in the future development of ChatGPT.