Release GPT-5.5 Instant as default model

## What changed OpenAI rolled out GPT-5.5 Instant as the default model across all ChatGPT users starting May 5, 2026. It replaces GPT-5.3 Instant and targets everyday prompts with gains in factual accuracy, image understanding, STEM handling, and web search results.

The new model produces tighter answers, cuts unnecessary follow-up questions, and drops overformatting plus extra emojis. Plus and Pro users gain better use of context from past chats, uploaded files, and connected Gmail accounts.

## Why it matters This shift prioritizes reliability and brevity for the majority of interactions instead of pushing users toward heavier models. It pressures older default options and signals OpenAI's bet that most builders want faster, less cluttered outputs without manual model switches.

Solo operators gain from reduced repetition and cleaner context carryover, but they lose easy access to the prior default once the three-month window closes for paid plans.

## How to use it GPT-5.5 Instant activates automatically for all accounts on web, iOS, and Android. Paid users reach model settings through the composer picker to keep GPT-5.3 Instant active for up to three months before retirement.

Storage and memory features remain available under the same plan limits listed in the May 14 update, with Free users capped at 500 MB.

## Watch for Confirm the bet if repeat-question rates drop and file-grounded answers improve measurably. The move breaks if users report higher hallucination rates on edge cases or if paid tiers see forced migration complaints. Expect OpenAI to retire GPT-5.3 Instant fully and expand similar defaults to agentic tools next.