# Message Deletion and Token Management Tools for Long Chats

**URL:** <https://community.openai.com/t/message-deletion-and-token-management-tools-for-long-chats/1291442>\
**Category:** Feature requests\
**Created:** [June 18, 2025, 3:04am UTC](https://community.openai.com/t/message-deletion-and-token-management-tools-for-long-chats/1291442 "2025-06-18T03:04:06Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![brownie2](https://avatars.discourse-cdn.com/v4/letter/b/94ad74/32.png) [@brownie2](https://community.openai.com/u/brownie2)\
**Post date:** [June 18, 2025, 3:04am UTC](https://community.openai.com/t/message-deletion-and-token-management-tools-for-long-chats/1291442/1 "2025-06-18T03:04:06Z")

</div>

As a long-form ChatGPT user, I frequently hit the token limit within a single chat, especially when working on creative writing, worldbuilding, or iterative projects. Unfortunately, there’s currently no way to delete individual messages — even outdated or redundant ones — to free up token space. This results in the need to start over in a new chat, breaking continuity and forcing manual copy-pasting of important context from earlier conversations.

I’d like to request the following features, all of which would dramatically improve workflow and usability for power users like myself:

1. **Delete a single message and its associated reply.**
2. **Delete a message and all subsequent messages (“cut from here down”).**
3. **An indicator of how much of the current conversation is taking up the active context window (tokens).**
4. **A way to collapse or summarize earlier messages to preserve context while trimming token cost.**

These features wouldn’t just improve efficiency — they’d remove major pain points for people using ChatGPT as a serious creative or productivity tool. Right now, it’s frustrating to lose critical context simply because we can’t curate our own message history.

This is especially frustrating because glitches sometimes _do_ remove user messages without warning (which in the past few days has happened to me over a dozen times), causing the model replies before and after that message to merge together into one message — proving the system _can_ alter conversation history, just not in a way the user controls. Giving users control over this process would make the platform far more useful and respectful of our time and work.

Thanks for considering this request. I believe implementing these tools would empower a wide segment of your dedicated user base. I love the platform and just want to see it become more functional for people doing deep, ongoing projects.