ChatGPT makes memory explicit

OpenAI now lists Memory as a visible ChatGPT capability, instead of leaving it as a mostly behind-the-scenes behavior. (help.openai.com) In OpenAI’s capabilities guide, Memory appears alongside tools such as web browsing, file uploads, data analysis, voice mode, and Canvas. The page says ChatGPT can remember facts such as a user’s name, preferences, or goals when memory is turned on. (help.openai.com) OpenAI’s memory documentation, updated in early April 2026, says the system works in two layers: “saved memories,” which users explicitly ask ChatGPT to keep, and “chat history,” which lets ChatGPT draw on past conversations. Users can turn on one, both, or neither in settings. (help.openai.com) The same help pages say Memory is for durable preferences and personal context, not for storing exact templates or long verbatim text. OpenAI also says users can ask what ChatGPT remembers, delete individual memories, clear all memories, or use Temporary Chat so nothing is referenced or saved. (help.openai.com) That shift lands as OpenAI also makes model choice more explicit inside ChatGPT. A help article updated about 10 days ago says GPT-5.3 Instant is now the default for logged-in users, while GPT-5.4 Thinking and GPT-5.4 Pro are presented as separate options for harder tasks. (help.openai.com) OpenAI describes GPT-5.4 Thinking as better at “keeping track of what it has already done,” which reduces how often users need to restate details inside a long task. That framing puts memory, model selection, and continuity in the same product surface instead of scattering them across separate settings and behaviors. (help.openai.com) Before these controls were easier to see, users often built their own continuity system with prompts. A Tom’s Guide write-up republished by MSN on April 12, 2026 described a “memory cheat code” that starts a chat by telling ChatGPT to treat key instructions as persistent context for the session. (msn.com) OpenAI’s own docs point to the same user need in a more formal way. The Projects feature, updated six days ago, is pitched as a workspace where chats, files, and instructions stay together so ChatGPT “remembers what matters and stays on-topic” over repeated work. (help.openai.com) The result is a clearer product line: one set of controls for what ChatGPT should remember about you across chats, another for what model should handle the task, and a separate Temporary Chat mode for conversations that should leave no memory behind. (help.openai.com)