The Codex "model full" error rate decreased from 6.82% to 0.085%
BIBIBI
AT A GLANCE
According to statistics cited in the report, the proportion of Codex users experiencing storage errors dropped from 6.82% on October 19 to 0.085% on November 12.
Article
Codex "Model Full" errors have been significantly reduced: the proportion of affected users has dropped from 6.82% to 0.085%
Dynamic Observation Beating AI news flash: on OpenAI, core product lead Tibo Sottiaux shared an internal statistic showing that the team may have finally resolved the long-standing "model capacity full" error plaguing users.
The data shows that on October 19, about 6.82% of Codex daily active users encountered at least one request that ended with a capacity error. By November 12, this proportion had fallen to 0.085%.
Previously, many users reported that even with sufficient subscription credits, Codex would suddenly prompt "model full," requiring them to switch models. Ongoing programming tasks would be abruptly interrupted, and switching models didn't always help.
This fix primarily focused on the automatic retry mechanism. October 15 to November 6,OpenAI made 6 rounds of adjustments in succession. When the server is overloaded, it tells Codex how long to wait before retrying. After receiving the instruction, the client can wait and automatically retry, minimizing direct task termination. OpenAI also increased inference compute power beforehand.
Some joked that the reduction in Codex errors might be because users have all switched to using Claude Opus 5.5.
The report said that Tibo Sottiaux, head of core products at OpenAI, shared an internal chart and stated that the team may have resolved the "model capacity exceeded" error.
02
The report cited statistics stating that on October 19, about 6.82% of Codex daily active users encountered at least one request that ended in a capacity error; by November 12, that proportion had dropped to 0.085%.
03
The report said that OpenAI conducted 6 rounds of automatic retry mechanism adjustments during October 15 to November 6.
04
The mechanism described in the report is: when the server is overloaded, it tells Codex how long to wait before retrying; after receiving the instruction, the client waits and automatically retries.
05
The report said that OpenAI had previously increased its reasoning compute.
06
Reports mention that people joked that the reduction in errors might be due to users switching to Claude Opus 5.5.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
The core of the report is that when the Codex encounters server capacity issues, the system adds a processing method of waiting and then automatically retrying, aiming to avoid directly interrupting programming tasks due to request failures. The proportion of error-prone users listed in the article has decreased, but this does not prove that all users' problems have been resolved.
Why it matters to readers
If the reported ratio and improvements continue, fewer capacity errors may mean that the Codex task is less often interrupted by such errors; the report does not provide a chart for verification.
Newbies using Codex can understand that when the server is busy, the system may wait before retrying, rather than ending the request immediately. The report still states that changing the model may not solve the problem, so it cannot be assumed that tasks will not be interrupted.
Risks and unknowns
The original text does not include internal statistical charts, and the scope and data of the statistics cannot be verified independently.
Reports say the team “may” have resolved the issue; it is not yet confirmed that all user capacity errors or task interruptions have been eliminated.
It is not specified in the original text whether the decrease in error ratio is caused solely by the auto-retry adjustment.
Related Developments
Loading event timeline…
Related concepts
Nikkei active users: Daily active users used for statistics reporting.
This term is not in the glossary yet. Browse related concepts in the glossary.
View glossary →Auto-retry: When a request fails or requires waiting, the client waits as indicated and tries again.
This term is not in the glossary yet. Browse related concepts in the glossary.