Otter Potter

Model Truth for ChatGPT

Picked a model, but did it answer? Model Truth puts a small label under every ChatGPT answer naming the model that wrote it, marks the ones that came from a different model than the rest of the chat, and counts how many answers each model wrote for you, since ChatGPT doesn't show how much you have left.

Free labels and 3-hour count. Pro is $4.99, paid once: 24-hour and 7-day history, and caps that turn the toolbar badge amber at 80%.

Three ChatGPT answers labelled GPT-5, GPT-5 mini (not GPT-5) in amber, and GPT-5.
The second answer came from a smaller model.

Questions

Where does the model name come from?
From the conversation data the ChatGPT page itself loads; each answer records the model that wrote it.
What does it collect?
Nothing leaves your browser. For the usage count it keeps the model name and time of each new answer, in your browser, for 7 days; no message text. See the privacy policy.
Is this made by OpenAI?
No. Model Truth is an independent extension by Otter Potter and is not affiliated with OpenAI.

Also for ChatGPT