OpenAI Releases GPT-5.5 Instant as New Default ChatGPT Model

OpenAI on May 5, 2026 began rolling out GPT-5.5 Instant, a new foundation model that replaces GPT-5.3 Instant as the default model in ChatGPT. According to TechCrunch's report and OpenAI's release notes, the company positions GPT-5.5 Instant as a step-up on factual reliability rather than a step-up on raw capability: OpenAI says the model reduces hallucination in sensitive areas such as law, medicine, and finance while preserving the low-latency profile that defines the Instant tier.

Around the model change, OpenAI shipped a set of consumer-facing changes that lean on it. ChatGPT added memory improvements for Plus and Pro users so responses can pull continuity from past chats, saved memories, and the connected Gmail app. The company also introduced Trusted Contact, a safety feature that lets users designate someone ChatGPT can reach in moments of acute distress, advanced its voice intelligence models in the API, and began testing ads inside ChatGPT. On the enterprise side, OpenAI announced expanded Trusted Access for Cyber on May 8 with GPT-5.5 and a specialized GPT-5.5-Cyber variant.

The release lands inside a busier model week than usual. Anthropic shipped Claude Opus 4.7 for financial work and signed a 300-megawatt compute deal with SpaceX. Google is widely expected to unveil major Gemini upgrades at I/O on May 19, 2026. Against that backdrop, OpenAI's choice to push a quieter Instant-tier refresh as the default model — rather than a flagship reasoning release — reads as a bet that hallucination on regulated questions is the friction that hurts ChatGPT most in everyday use, and that fixing it is a more defensible product win than a benchmark leap that only power users notice.