Question on Subscription Quota Parity Between Interactive Codex TUI, codex exec, and codex app-server

[jyongchul]

Request for Clear Documentation on CLI & codex app-server Subscription Metering

In light of recent undisclosed quota consumption differences between interactive TUI and non-interactive (-p / SDK) modes in other coding CLIs (see anthropics/claude-code#107106, #107134, and google-antigravity/antigravity-cli#1411), we would appreciate official clarification on how openai/codex meters subscription usage across entrypoints:

1. Third-party coding environments such as pingdotgg/t3code use existing codex login subscriptions over codex app-server stdio JSON-RPC (docs/user/faq.md L15–L24, L56–L60), while noting that on other providers non-interactive SDK entrypoints are metered ~1.8× faster than interactive CLI unless a custom entrypoint is configured (ClaudeHome.ts L66–L88, youtu.be/y5TEATUDDfc at 12:00–15:01).
2. Could the Codex team explicitly document in the repository README or FAQ whether interactive codex TUI, non-interactive codex exec, and codex app-server consume subscription quota at identical 1:1 rates, and commit to advance notice before introducing any mode-specific quota multipliers?

[jyongchul]

Follow-up: Publicly documented CLIProxyAPI hub setup (docs/user/faq.md lines 58–60 & usage.md lines 176–186) and request for clarification under OpenAI’s Terms of Use

Following up on the initial discussion post above (#59674), we are sharing additional primary-source references from pingdotgg/t3code (commit 337c5858c3), an unlisted Gist linked from its public FAQ (gist.github.com/t3dotgg/59a521d2ee8d70aaf90e280c175072af), and a companion video linked in the same FAQ (youtube.com/watch?v=D8PikZ1KhUo) so that the Codex team can clarify the permitted boundaries under OpenAI’s Terms of Use.

1. Contrast between docs/user/faq.md line 58 and line 60 (CLIProxyAPI Codex hub guide & UI integration)

• In docs/user/faq.md (lines 50–58), the project's documentation states:

  "Will I get banned for using my subscription? No. We work with all of the labs and T3 Code is within the TOS of Claude Code, Codex, and Google Antigravity... If it works in T3 Code, you can trust it is safe."
  "I don't think T3 Code is the right place to manage a pile of accounts. If we add an auto-switcher, we are risking TOS bans for our users."

• Immediately below on line 60 of the same FAQ (and in docs/user/usage.md lines 176–186), however, the documentation links directly to an external proxy setup and built-in hub integration:

  "... For a bigger pool, manage them outside T3 Code with a self-hosted proxy like CLIProxyAPI. I walked through how I use 6 Claude subs and 3 Codex subs in this video, and wrote up the setup in this guide. Hand it to your agent and it will walk you through it. Early apologies if it gets you banned 💀 After you set up a hub, T3 Code can show its pooled limits under Usage. Connect a hub"

• The linked guide (an unlisted Gist linked from the project's public FAQ, cliproxy-setup.md, titled "How I manage 7 Claude and 3 Codex accounts across 12+ computers") describes signing several personal subscriptions into one machine running CLIProxyAPI and pointing other machines at that single endpoint (model_provider = "cliproxy", requires_openai_auth = false), and itself states:

  "This might get you banned. This might be against the terms of service." (line 82)

• In docs/user/usage.md (lines 176–186), the documentation also describes connecting the proxy hub into the application UI:

  "To see pooled accounts, open Settings → Providers → Usage providers → Add hub... The accounts appear under Usage → Limits. Codex accounts show banked reset credits; select an account and choose Use reset to redeem one."

2. Relationship to OpenAI’s Terms of Use & impact on developers

OpenAI’s Terms of Use (What You Cannot Do) prohibit actions that:

"Interfere with or disrupt our Services, including circumvent any rate limits or restrictions or bypass any protective measures or safety mitigations we put on our Services."

We are not in a position to judge the setup described in the linked Gist; we only note that the same FAQ says built-in account switching risks bans (line 58) while linking that guide and a hub integration (line 60; usage.md lines 176–186). When setups like this are shared publicly with agent-ready instructions, providers may respond with stricter limits for everyone. Developers who follow the official documentation should not be the ones who absorb that cost, which is why we ask for one published rule that applies equally to every customer and tool.

3. Questions for the OpenAI Codex team

1. Official stance under the Terms of Use: We would like to know whether the multi-subscription proxy setup linked from docs/user/faq.md (line 60) and integrated via docs/user/usage.md (lines 176–186) is permitted under OpenAI’s Terms of Use, or whether it falls under "circumvent any rate limits or restrictions". Whatever the answer, we ask that OpenAI publish it clearly as a general rule.
2. Clear, uniform metering documentation: Will OpenAI publish explicit guidance on the permitted boundaries and quota consumption rules across codex interactive TUI, codex exec, and codex app-server so that all developers have transparent, predictable terms?