Regression: Token usage significantly increased in v0.121.0 vs v0.116.0 for identical prompt and unchanged workspace
What issue are you seeing?
After upgrading OpenAI Codex from v0.116.0 to v0.121.0, token usage increased significantly (by ~60%+) for the exact same prompt, executed in the same workspace, with no code or configuration changes.
Interestingly, v0.116.0 appears to be the most efficient version, while both v0.115.0 and v0.121.0 use substantially more tokens. This suggests a regression in context handling introduced after v0.116.0.
Environment
Codex versions tested:
v0.115.0
v0.116.0
v0.121.0
Model: gpt-5.3-codex medium
OS: Windows
Workspace: same project, no changes between runs
Prompt: identical in all cases
What steps can reproduce the bug?
Open the same project workspace
Run Codex with version v0.115.0
Ask: Please summary me the code within the workspace
Note token usage
Repeat with:
v0.116.0
v0.121.0
Ensure:
no changes to code
no changes to configuration
same prompt used
What is the expected behavior?
Token usage should remain relatively stable across versions for the same prompt and workspace, or change only slightly due to minor improvements.
Additional information
Actual Behavior
Token usage varies significantly between versions:
v0.115.0
total=63,395
input=60,290 (+570,368 cached)
output=3,105
reasoning=813
v0.116.0 ✅ (most efficient)
total=52,896
input=50,392 (+333,056 cached)
output=2,504
reasoning=548
v0.121.0 ❌ (regression)
total=86,194
input=83,231 (+673,664 cached)
output=2,963
reasoning=1,040---
Summary
- v0.116.0 → best efficiency (lowest input, cached, and reasoning)
- v0.115.0 → moderate usage
- v0.121.0 → clear regression (highest across all metrics)
6 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
This is expected. Newer versions include more built-in tools. You should expect this number to continue to rise as we add more capabilities to Codex.
Ok, so in that case, the limits in the paid plans as well are considered to rise?
Thanks to the entire codex team for the work done.
Regarding the token usage, I've experienced the same situation--no new features were being used, so increased token consumption was not expected.
@fipciu1996 Thanks for your precise version analysis. I've downgraded to 0.116 for now, which has reduced token consumption.
Just to clarify, we do try very hard to keep the base token count to a minimum. We'll continue to make that a priority. However, new functionality sometimes requires us to add more built-in tools and extensions to the base prompt.
@etraut-openai
The main concern is that, in practice, end users will end up paying more. The current “Plus” subscription no longer seems sufficient to cover the increased token usage introduced by the new functionalities.
It appears that new features are being added without a proportional increase in token limits, which effectively shifts the cost burden to the user.