OpenAI's next-generation model GPT-5.6 is showing up inside ChatGPT Pro before the company has formally announced or released it, according to multiple user reports over the past 48 hours.

Early testers say the new model improves reasoning speed and factual accuracy over the current version, though it still exhibits occasional errors on complex coding tasks. The leak has reignited debate about how OpenAI manages staged rollouts and whether it can control access to proprietary systems once deployed to paying customers.

A spokesperson for OpenAI said the company is closely monitoring the situation and did not confirm whether the model was intended for general release this quarter. The model's appearance in Pro accounts suggests OpenAI may be performing an A/B test or a soft rollout without public announcement.

GPT-5.6's partial access arrives at a sensitive moment for ChatGPT. Consumer surveys published last week placed ChatGPT's share of the consumer AI market below 50 percent for the first time, with competitors such as Anthropic and Google closing the gap on enterprise clients.

Analysts say that leaking a premium model into production accounts, even inadvertently, risks fragmenting the user experience. Customers paying for Pro access expect consistent performance, not experiments that may be withdrawn or altered without warning.

OpenAI has not set a timeline for a full GPT-5.6 launch, but the company typically reveals major model upgrades at scheduled developer events. The current leak has left the industry speculating that an announcement could come as early as next week.

Image source: v3b.fal.media