TechCrunch: OpenAI releases GPT-5.5 Instant as the new default ChatGPT model

On May 5, 2026, OpenAI released GPT-5.5 Instant as the new default model for ChatGPT, replacing GPT-5.3 Instant. The update is relevant to writers and journalists who use ChatGPT for research, drafting, and editing assistance.

The most significant change for content work is reduced hallucination in sensitive topic areas. The model performs more reliably in domains where factual accuracy carries high stakes — law, medicine, and finance — which matters for journalists, technical writers, and content teams working with regulated or complex subject matter. OpenAI does not claim hallucination is eliminated, but the improvement is measurable and documented in benchmark comparisons.

The model now accesses past conversations, uploaded files, and Gmail to provide context-aware responses. For writers who maintain long-running projects in ChatGPT — research threads, editorial style references, ongoing series — this means the model can draw on accumulated context without manual re-prompting. A companion update makes memory sources visible across all models: the system now displays where it is drawing information from, giving users the ability to verify and correct the sources the model is using.

GPT-5.5 Instant scored 81.2 on the AIME 2025 benchmark, up from 65.4 on the previous version, and 76 on MMMU-Pro. Stronger reasoning is most directly useful for analytical writing tasks — summarizing contradictory sources, constructing arguments from evidence, identifying gaps in existing coverage.

For developers building AI-assisted writing tools, the model is available through the API as “chat-latest.” Teams that have built workflows on top of GPT-5.3 Instant will receive the upgrade automatically under that alias.