1. Computer History Creates Context
The most significant change is the introduction of 'Computer History', a new feature in the desktop app. Once you opt-in, ChatGPT can create a timeline of your activity across selected applications and websites. This doesn't involve screenshots but allows
the AI to understand the context of your work without you needing to explain it repeatedly. For example, if you spend the morning researching a topic across multiple browser tabs and documents, you can later ask ChatGPT to summarize your research, and it will know which sources you consulted. You have full control, with the ability to pause the feature, exclude apps, and delete your history at any time.
2. Agentic, Web-Aware Workflows
ChatGPT is now capable of taking action on your behalf across the web. The latest updates allow the 'Work' version of ChatGPT to log into websites securely to complete tasks. It can handle everything from booking an appointment to figuring out utility plans for a new home and signing up. This is enhanced by new webhook-triggered tasks for Plus and Pro users. You can now prompt ChatGPT to act based on events in other applications, such as summarizing an email the moment it arrives in Gmail or drafting a response when a specific Slack message is posted. This shifts the AI from a passive assistant to a proactive agent that automates multi-step processes.
3. Enhanced and Partitioned Memory
The Memory feature, which allows ChatGPT to remember facts about you and your preferences across conversations, has been significantly improved. The system is now more nuanced, with a feature called 'Projects' that acts as a self-contained workspace. Each project has its own partitioned memory, meaning what ChatGPT learns about a specific project stays within that context and doesn't bleed into your general chats or other projects. This allows for deep, specialized knowledge for individual workstreams without confusing the AI's general understanding of your preferences. You can view, edit, and delete what the model remembers, ensuring you remain in control of your data.
4. Smarter Voice and Broader Access
Voice interactions continue to become more natural and capable. The latest updates allow you to talk through files and projects with ChatGPT Voice, making it a more versatile collaborator. This is part of a broader push for accessibility. OpenAI has removed the message limits for text chats on the free tier, although caps remain for more intensive tasks like image generation and file uploads. Furthermore, the desktop app has expanded its reach, now supporting Linux and offering better integration with browsers like Edge, Brave, and Opera, allowing you to bring an open tab directly into a chat.
5. New Models and More Direct Links
Underpinning these new features is a new family of models, including the GPT-5.6 series. Different versions are available depending on your subscription plan, with the most powerful, 'GPT-5.6 Sol', being reserved for complex reasoning tasks on paid plans. For users on the free tier, 'GPT-5.6 Luna' is the default model. Alongside model improvements, ChatGPT is also becoming a more significant source of web traffic. Recent changes show the AI is linking out directly to external websites more often in its responses, providing measurable referral traffic to content creators and businesses. This marks a shift from being a closed answer engine to a platform for discovery.
