Recently, ChatGPT has made “more usable” concrete in two small ways: first, it can remember you at the right times; second, it lets you pull up old conversations faster. For people who write, study, or run projects day to day, these updates feel more tangible than simply stacking model parameters.
ChatGPT Memory: from “remembering” to “controlled remembering”
ChatGPT’s memory is no longer just a gimmick—it can retain the preferences you mention repeatedly, reducing the time you spend re-explaining context each time. More importantly, ChatGPT provides a management entry point: you can see what it has remembered, delete a specific memory, or turn off memory with one click.
If you only want to ask sensitive or one-off questions temporarily, you can use Temporary Chat to avoid memory and logging. That way, ChatGPT can both “understand you better the more you chat” and “lose its memory on the spot” when you need it to.
Conversation Search is here: no more digging through history for old content
Many people use ChatGPT as a workbench; after a while, the conversation list can get very long. Now you can search your conversation history directly in ChatGPT and quickly locate a set of meeting notes, an explanation of a piece of code, or a writing outline by keyword.
Even more useful is that the “cost of continuing a conversation” is lower: once you find an old thread and keep asking questions there, ChatGPT can reuse the context from that time, so you don’t have to paste the materials all over again.
