【How to Resolve 401 After Upgrading to Codex 0.149.0】

If you encounter API_KEY_REQUIRED / 401 Unauthorized after the upgrade, it is because the new version no longer allows custom Providers to automatically inherit authentication from auth.json when requires_openai_auth=false.

Solution:

Manually modify the configuration file Please modify the configuration file:

macOS: ~/.codex/config.toml Windows: %USERPROFILE%\.codex\config.toml

Find this line: requires_openai_auth = true (change the original false to true to use it normally)

The deepseek-v4-flash channel is temporarily shut down. We are looking for an alternative solution and have added access to muse-spark-1.2 in the check-in quota pack, which can be used as a temporary replacement.

It has now recovered~~OpenAI was malfunctioning, and most requests were returning this error. We need to wait for the official fix. The current mitigation is to start a new conversation, try passing the old conversation's id to Codex, and ask Codex to read the conversation information and continue: status_code=400, Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt:

Recently, openai modified the cache hit rules, resulting in a decline in the current cache hit rate. We are trying to optimize the scheduling logic to better match the updated cache hit rules.

Standard response: if you think it’s laggy, it’s an openai issue: {"error":{"type":"service_unavailable_error","code":"server_is_overloaded","message":"Our servers are currently overloaded. Please try again later.","param":null}}