Luna usage draining codex usage at the same speed as Sol despite no difference in the way it's used

Open 💬 3 comments Opened Aug 3, 2026 by antoniochrist42005-rgb
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of the Codex App are you using (From “About Codex” dialog)?

26.802.11031 (newest possible windows version)

What subscription do you have?

Plus

What platform is your computer?

Windows 11

What issue are you seeing?

Luna max uses drains codex limits at a similar rate to Sol high, despite having 25x lower API pricing. I use Luna in the same coding environement, no fast mode, no subagents and similar tasks that also get completed successfully, however Luna is draining my limit at Sol speed. I suspect a usage billing issue. Please investigate this blatent overcharging.

What steps can reproduce the bug?

Using the Luna model

What is the expected behavior?

Luna draining my usage comparable to Sol, even though it should drain it 25x slower per token.

Additional information

Same coding environment, no fast mode, no subagents, no infinite stuck loops (you guys ruined my weekly usage this week with this)

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 24 days ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

  • #36144

Powered by Codex Action

bombaris34 · 24 days ago

It seems it sometimes does not switch the model correctly

antoniochrist42005-rgb · 24 days ago
It seems it sometimes does not switch the model correctly

This is a consistant thing that happens every single time and i have already checked which model tokens are being used. It is Luna, not Sol but the usage still drains like Sol. It is a massive shame because i am the exact type of person that really needs a cheap model as capable as Luna without needing to pay API prices for deepseek. I am the target audience.