• Pro history is filled in from your recent chats the moment you unlock it, instead of starting from zero.
• Caps can be set before a model has answered, with rounded starting points for your plan that you can edit.
• Pro turning on is visible: the toolbar shows PRO and the popup says so.
• Answers ChatGPT recorded no model for now read "Model not recorded" instead of showing nothing.
• Model names follow OpenAI's current lineup and capitalisation, e.g. "GPT-5.4 Thinking", "GPT-5 mini".
• If ChatGPT changes something that stops models being read, Model Truth now tells you instead of going quiet.
• A purchase is no longer missed when checkout is slow, and Pro is never switched off by a licence check that couldn't be made.
Your counts, caps and Pro status carry over.