CHATGPT SWAPS DEFAULT MODEL FOR FACTUAL GPT-5.5

OpenAI has begun deploying GPT-5.5 Instant to all ChatGPT users, replacing the previous default model with a version designed to deliver more accurate and personalized responses. ■ Fewer Hallucinations Internal testing shows GPT-5.5 Instant produced 52.5 percent fewer hallucinated claims on sensitive topics including medicine and law. This improvement addresses a persistent challenge with large language models—generating confident-sounding but factually incorrect information. ■ Memory Sources Feature A new feature called "memory sources" lets users see which stored context shaped a given response. This transparency mechanism helps users understand how ChatGPT arrived at its answers, particularly useful when the model draws from past conversations, uploaded files, or other personalized data. ■ Rollout Timeline GPT-5.5 Instant is available to all ChatGPT users immediately as the default model. Personalization capabilities—which leverage past chats, files, and Gmail integration—are launching first for ChatGPT Plus and Pro subscribers on the web. Broader rollout of personalization features to other user tiers and platforms will follow. ■ Impact The shift reflects OpenAI's focus on reducing AI model unreliability while expanding personalization. For users reliant on ChatGPT for factual information, the hallucination reduction could significantly improve trust. For those using the service with personal context, the memory sources feature provides greater control and visibility over how their data influences responses. OpenAI has not specified when GPT-5.5 Instant personalization will reach free tier users or mobile platforms.