Flash News from Beat AI, OpenAI Codex lead Tibo Sottiaux responds to the recent quota reduction controversy. He stated that OpenAI will not secretly modify quotas. After the team investigated some affected users, they found that many were using sub2api. Converting subscriptions to API traffic and then distributing or sharing it among multiple people would trigger OpenAI's anti-fraud system.
Ironically, Tibo recommended another proxy tool, CLIProxyAPI, just a month ago. He taught users to log in with Codex OAuth and then integrate GPT-5.6 Sol into Claude Code. At the time, he even joked, "If this trick gets blocked, consider it a free reset from me."
CLIProxyAPI and sub2api are not the same project, but their functions are not that different. CLIProxyAPI can also integrate Codex OAuth with compatible APIs, and it supports Codex multi-account polling. On the other hand, sub2api goes a step further, specifically offering multi-account management, API key distribution, billing, and multi-user sharing.
Therefore, the red line drawn by OpenAI now seems more like "individual proxy is acceptable, but multi-user distribution is not." The issue is that just a month ago, the lead was publicly teaching users to go through a proxy, and now they are explaining that some usage scenarios will trigger risk control. This boundary was not clear before.
Moreover, this still does not explain all the anomalies. In Tibo's comments section, there are also users who only used the official Codex and claim a significant quota reduction. Currently, this round of controversy cannot be entirely attributed to sub2api.
