ChatGPT Release Notes Detail Enhanced App Experience and Memory Features
OpenAI has recently rolled out a series of updates to ChatGPT, focusing on improving user experience and expanding its capabilities. The official release notes highlight several key changes, many of which are designed to make the AI assistant more intuitive and efficient for a wider range of users.
One notable improvement addresses the handling of large text inputs. For ChatGPT Free and Go users, any text paste exceeding 10,000 characters will now be automatically converted into an attachment. This change aims to maintain a cleaner composer interface and prevent extensive text from consuming the entire context window, a feature already available to Plus, Pro, and Business subscribers. Users retain the flexibility to convert these attachments back into inline text if needed.
The ChatGPT application across web, iOS, and Android platforms has also received substantial enhancements. These app experience updates, initially rolled out around June 18, 2026, introduce practical new functionalities. Users can now benefit from pronunciation guidance for over 60 languages, available in both audio and text formats, making language learning and communication more accessible. A dedicated feature for World Cup updates allows for conversational tracking of the tournament, covering schedules, matchups, and predictions. Additionally, chat organization and sharing have been streamlined, and users now have more control over app permissions, choosing when ChatGPT prompts for consent before using connected applications. iOS users will also experience faster photo uploads.
A significant development is the expansion of ChatGPT's memory system, referred to as "dreaming." This advanced memory synthesis system is now being rolled out to Free users, having previously been available to Plus and Pro subscribers. The "dreaming" feature allows ChatGPT to automatically curate and update memories in the background by referencing chat history, ensuring that past interactions and preferences are more accurately and relevantly recalled in future conversations. This update aims to provide a more personalized and continuous user experience, with the rollout to Free and Go plans, as well as additional countries, expected in the coming weeks.
In terms of underlying models, GPT-5.3 Instant Mini has been deployed as a fallback model for users who reach their rate limits for GPT-5.3 Instant. This mini version is designed to offer more natural conversations with improved writing and contextual awareness compared to its predecessor, GPT-5 Instant Mini. Concurrently, OpenAI has announced the retirement of older GPT-5.2 models, including GPT-5.2 Instant, Thinking, and Pro, as part of an ongoing effort to streamline and focus on its latest, more performant models. Existing conversations that utilized GPT-5.2 will automatically transition to the corresponding GPT-5.5 model.
Read original source