Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
OpenAI ChatGPT is rolling out a new model today, with a focus on making output more useful, personal, and accurate. According to OpenAI, GPT-5.5 Instant will provide clearer, concise, and less overly formatted answers to everyday queries. The new form will limit follow-up questions and “gratuitous emojis” to make responses less chaotic.
The developer says GPT-5.3 Instant hallucinates significantly less than its predecessor on “high-stakes prompts” across subjects such as finance, law and medicine, and performs better across picture reasoning, science and mathematics.
Read also: OpenAI has created a special ChatGPT for your doctor
Improved memory allocation and control
OpenAI is also tipped to improve personalization, saying that GPT-5.5 Instant can use chats, files, and previously connected Gmail accounts to answer prompts tailored to you, your workflow, and your dialect.
The company is also rolling out new memory resources, giving you more control, context, and visibility into what information ChatGPT uses to generate personalized responses.
“When a response is personalized, you can see what context it was used in, such as saved memories or previous chats, and delete or correct it if something is outdated or no longer relevant,” OpenAI’s announcement said.
“Memory sources are not visible to others if you choose to share a conversation. You will remain in control of what is in your memory: you can delete chats that you no longer want to recall, delete items in saved memories or change them in settings, or use or update temporary chats that do not use your memory.”
GPT-5.5 Instant is available on Tuesday, replacing GPT-5.3 Instant as the default model in ChatGPT and in the API as “Latest Chat.” GPT-5.3 Instant will remain available for another three months to paid users, after which it will be retired, OpenAI said.
Read also: ChatGPT 5.5 is about math, science, and AI research