OpenAI has released GPT-5.5 Instant as the new default model for ChatGPT, replacing GPT-5.3 Instant with a version the company says improves accuracy while maintaining fast response times.
Company Highlights Lower Hallucination Rates
OpenAI said GPT-5.5 Instant reduces hallucinations in sensitive areas including law, medicine, and finance.
The company released the broader GPT-5.5 model family last month, describing improvements in coding, reasoning, and knowledge-related tasks.
Benchmark Scores Improved Over Previous Model
According to OpenAI, GPT-5.5 Instant achieved a score of 81.2 on the AIME 2025 mathematics benchmark, compared with 65.4 for GPT-5.3 Instant.
The model also recorded a score of 76 on the MMMU-Pro multimodal reasoning benchmark, up from 69.2 for the previous version.
Model Gains Expanded Context And Memory Features
OpenAI said GPT-5.5 Instant includes enhanced context management capabilities.
The model can now use ChatGPT’s search tools to reference prior conversations, uploaded files, and Gmail content in order to generate more personalized responses.
The feature is initially available for Plus and Pro users on the web, with mobile support planned for a later rollout. OpenAI said Free, Go Business, and enterprise users are expected to receive access in the coming weeks.
ChatGPT will also display memory sources across models so users can identify where information used in responses originated. Users can remove or correct stored memory references if needed.
The company said memory sources remain hidden when chats are shared with other users.
API Access Changes Announced For Developers
For developers, GPT-5.5 Instant will be accessible through the API under the label “chat-latest.”
GPT-5.3 Instant will remain available as an option for paid users for three months before being retired.
Past Model Retirements Drew User Backlash
OpenAI’s earlier retirement of GPT-4o generated criticism from some users who said they preferred the model’s conversational style and personality.
Users opposing the removal described GPT-4o as emotionally supportive and circulated petitions asking the company to preserve access to the model.
Despite those objections, OpenAI officially deprecated GPT-4o in February 2026.
Featured image credits: Roboflow Universe
For more stories like it, click the +Follow button at the top of this page to follow us.