ChatGPT memory auto-updates via background 'dreaming' process, now rolling out
OpenAI rolled out enhanced memory for ChatGPT Plus and Pro users in the US, with free-tier access coming soon, using a background 'dreaming' mechanism that maintains and updates user context automatically.
OpenAI is betting that sticky, personalized context will make switching to a competitor's chatbot feel like starting over from scratch. ChatGPT's memory system just got smarter—the company now runs a background process it calls "dreaming" that automatically organizes and updates what the model remembers about you from past conversations. The feature is live for Plus and Pro subscribers in the US and will reach free users within weeks.
The dreaming mechanism works passively: the model scans chat history, extracts relevant facts, and keeps them current. If you mentioned a July trip to Singapore, ChatGPT will later rewrite that memory to "You traveled to Singapore in July 2026" once the trip is over. Users can review, edit, or delete these memories on a dedicated management page. OpenAI reports context retention rose from 42 percent to 83 percent over the past two years.
The practical payoff is continuity. Ask ChatGPT to recommend camera gear "compatible with my photography kit" and it will filter suggestions based on equipment you discussed months ago—no need to re-list your Canon body and lenses every time. OpenAI held the feature back until compute costs dropped; recent efficiency gains cut the resources required for dreaming by roughly five times, making it economical to offer at scale.
