OpenAI launches GPT-5.5 Instant as ChatGPT's default, cutting hallucinations by 52.5 percent on high-risk prompts

OpenAI is replacing GPT-5.3 Instant with GPT-5.5 Instant as ChatGPT's default model. In internal testing, GPT-5.5 Instant produced 52.5 percent fewer hallucinated claims than its predecessor on high-risk prompts in medicine, law, and finance. On conversations previously flagged for factual errors, inaccurate claims dropped by 37.3 percent.

GPT-5.5 Instant scores higher on multiple benchmarks: AIME 2025 accuracy jumped from 65.4 to 81.2 percent; GPQA (PhD-level science reasoning) rose from 78.5 to 85.6 percent; CharXiv (scientific chart reasoning) improved from 75.0 to 81.6 percent; MMMU-Pro (expert multimodal reasoning) climbed from 69.2 to 76.0 percent; OmniDocBench (document parsing) error rate dropped from 14.6 to 12.5 percent.

The model delivers shorter, less verbose answers without losing substance, asks fewer unnecessary follow-ups, and avoids superfluous emojis and heavy formatting. A new feature called 'memory sources' shows users which stored context—such as saved notes, past chats, or connected Gmail accounts—shaped a given reply; entries can be flagged, edited, or deleted.

GPT-5.5 Instant is rolling out to all ChatGPT users immediately. Paying users can access GPT-5.3 Instant through model settings for another three months before retirement. Enhanced personalization launches first for Plus and Pro users on the web, with Free, Go, Business, and Enterprise plans expected to gain access over the coming weeks.