Context window usage jumps to ~98% after minimal usage in new chats

What version of Codex is running?

Codex version: VS Code extension v0.4.62

What subscription do you have?

Plus

Which model were you using?

gpt-5.2-Codex

What platform is your computer?

Darwin 25.2.0 arm64 arm

What terminal emulator and version are you using (if applicable)?

VSCode with "Codex – OpenAI’s coding agent"

What issue are you seeing?

The context window usage indicator rapidly fills up to ~98% after very limited interaction. This occurs even in newly created chats with only one short prompt.

Example:
A chat containing only the question:

“what is 2+2 equal to?”

already shows the context window at ~98% usage.

The usage ramped up as well but I cannot quantify it as I would finish the remaining trying.

What steps can reproduce the bug?

Context window / token limit usage (as shown in UI)
• Fresh chat, single prompt.
• Context indicator shows approximately:
Context remaining: 98% left (2% used) 254k / 258k tokens left
(see attached screenshot)

Uploaded thread: 019be6d6-69c8-7361-9d77-dee4e16af6bd

What is the expected behavior?

new chats should start from 0% context and after one short prompt/response, context usage should remain low (far from ~98% used).

Additional information

<img width="297" height="757" alt="Image" src="https://github.com/user-attachments/assets/ba1bba18-2b3f-47b1-8bad-6313d43ee9ab" />