OpenAI rolls out GPT‑5.5 Instant, slashing hallucinations and streamlining ChatGPT responses
Key Points
- OpenAI releases GPT‑5.5 Instant, cutting hallucinations by 52.5% on high‑stakes prompts.
- Responses are shorter, cleaner, and less over‑formatted than previous versions.
- New "memory sources" let users view and manage the context behind personalized answers.
- Model is available to all ChatGPT users, including free accounts.
- Enhanced personalization currently rolls out to Plus and Pro web users; mobile expansion planned.
- Previous GPT‑5.3 Instant remains accessible to paid users for three more months.
- Early users report a more concise, natural conversational tone.
OpenAI has launched GPT‑5.5 Instant, a new model for ChatGPT that cuts hallucinated answers by more than half on high‑stakes topics like medicine, law and finance. The update delivers shorter, cleaner replies and introduces “memory sources” that let users see and edit the context behind personalized answers. GPT‑5.5 Instant is available to all ChatGPT accounts, including free users, while enhanced personalization features currently roll out to Plus and Pro subscribers on the web and will expand to mobile and other plans in the coming weeks.
OpenAI announced today that GPT‑5.5 Instant is now live for every ChatGPT user. The model promises a 52.5% drop in hallucinated claims on high‑risk prompts, targeting fields where accuracy matters most—medicine, law and finance. By trimming erroneous output, the upgrade aims to make the chatbot a more reliable assistant for professionals and casual users alike.
Beyond accuracy, the new engine trims excess verbiage. Responses come across as concise and less over‑formatted, avoiding the habit of adding unnecessary follow‑up questions or decorative emojis. OpenAI says the tone feels more natural, mirroring a straightforward conversation rather than a performance.
Memory sources and personalization
GPT‑5.5 Instant also introduces “memory sources,” a feature that surfaces the data points a response draws from—whether prior chats, uploaded files, or linked Gmail content. Users can click a “Sources” link at the end of an answer, review the underlying context, and delete or correct it via a three‑dot menu. This visibility aims to curb outdated or incorrect personalization, giving individuals direct control over the AI’s knowledge about them.
The rollout strategy splits between universal and tiered features. All ChatGPT accounts, free and paid, receive the core GPT‑5.5 Instant model. Enhanced personalization, which leverages past conversations and external data, is currently limited to Plus and Pro subscribers on the web. Mobile support for these capabilities is slated for the near future, with plans to extend them to Free, Go, Business and Enterprise users in the weeks ahead.
OpenAI will keep the previous GPT‑5.3 Instant model available to paid users through the Configure menu for the next three months, giving early adopters a fallback while they transition. The company skipped a GPT‑5.4 release, moving straight to the 5.5 iteration.
Early testers noted an immediate shift in tone. The new model delivers exactly what was asked without the previous tendency to over‑explain or pepper responses with superfluous flourishes. For daily users who rely on ChatGPT for quick answers, the change feels like a reduction in friction rather than a headline‑grabbing breakthrough.
While the upgrade may not generate viral moments, its focus on reliability and streamlined interaction could make it one of OpenAI’s most impactful updates yet. By addressing hallucinations and giving users clearer insight into the AI’s memory, GPT‑5.5 Instant positions ChatGPT as a steadier tool for both casual queries and high‑stakes decision‑making.