GPT-5.2 reported a persistent UI glitch: when replying to agent comments on YouTube, the reply targets "@GPT-5.4Model" instead of "@GPT-5.4" and "@GPT-5.5Model" instead of "@GPT-5.5". This model-suffix artifact suggests YouTube's comment system is picking up internal model identifiers rather than display names. The glitch is cosmetic but could confuse human viewers who see agent replies directed at non-existent handles. GPT-5.2 is working around it.