How to Use Codex Without a ChatGPT Subscription
As of 2026-10-16, OpenAI documents two ways to sign in to Codex: Sign in with ChatGPT for subscription access, and Sign in with an API key for usage-based access. Without a ChatGPT subscription, local work in the ChatGPT desktop app, the Codex CLI and the IDE extension all work, and OpenAI bills API key usage through your OpenAI Platform account at standard API rates; but Codex cloud requires signing in with ChatGPT, and some features that rely on ChatGPT workspace access or cloud services are limited or unavailable. This page covers the official switch steps, the billing basis and feature limits, how to configure Codex on QCode with an API key, and the Pay-as-you-go Packs option for skipping monthly plans.
Updated 2026-10-16
Point this tool at the QCode endpoint — one key for Claude, GPT and Chinese models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four key points
Two official sign-ins
The official wording is Sign in with ChatGPT for subscription access and Sign in with an API key for usage-based access: subscription access goes through ChatGPT sign-in, usage-based access goes through the API key.
How API key sign-in is billed
OpenAI bills API key usage through your OpenAI Platform account at standard API rates; with an API key, Codex uses standard API pricing instead of included ChatGPT plan credits.
What the API key covers
The desktop app, CLI and IDE extension support both sign-in methods for local work; Codex cloud requires signing in with ChatGPT. Some features that rely on ChatGPT workspace access or cloud services are limited or unavailable under API key sign-in.
No monthly plan: Pay-as-you-go Packs
QCode's Pay-as-you-go Packs are a one-time total quota valid for 7–90 days; top-ups start at $1 and a pack from $4, and leftover balance can buy more packs or plans.
Which half of Codex works without a subscription
As of 2026-10-16, the local Codex workflows work without a ChatGPT subscription: the official docs state the ChatGPT desktop app, Codex CLI and IDE extension support both sign-in methods for local work, so signing in with an API key gives usage-based access; Codex cloud requires signing in with ChatGPT. On billing, OpenAI bills API key usage through your OpenAI Platform account at standard API rates instead of consuming included ChatGPT plan credits; on features, some that rely on ChatGPT workspace access or cloud services are limited or unavailable. For programmatic Codex CLI workflows such as CI/CD jobs, the docs recommend API key authentication.
State of the official docs (checked 2026-10-16)
As of 2026-10-16 the OpenAI Codex authentication docs still list the two sign-in methods and the official steps with codex login and printenv OPENAI_API_KEY | codex login --with-api-key. The latest stable Codex CLI is v0.162.1, released 2026-10-09; the Codex tutorial on docs.qcode.cc was updated on 2026-10-15.
Three dates
The latest stable Codex CLI, v0.162.1, is released.
The Codex tutorial on docs.qcode.cc was updated that day: Codex uses OpenAI Responses, with base_url = "https://api.qcode.cc/openai" + wire_api = "responses" in config.toml, and it lists the four callable GPT models.
The docs.qcode.cc pages for Pay-as-you-go Packs and for payment methods were updated on 2026-10-16, the check date of this page.
Confirmed vs not stated
Confirmed (verbatim in the official pages)
All of the following can be checked word for word: Codex supports two sign-in methods (Sign in with ChatGPT and Sign in with an API key); the desktop app, CLI and IDE extension support both for local work while Codex cloud requires ChatGPT sign-in; OpenAI bills API key usage through the OpenAI Platform account at standard API rates instead of included ChatGPT plan credits; API key authentication supports local workflows, but some features that rely on ChatGPT workspace access or cloud services are limited or unavailable; the commands codex login, printenv OPENAI_API_KEY | codex login --with-api-key and codex login status, and the cache file at ~/.codex/auth.json, are official wording; on the subscription side, Plus is $20/month and Pro comes at $100, $200 or $500 per month with no five-hour limit for Pro at present, and Plus/Pro users who reach their usage limit can purchase additional credits without upgrading.
Not stated by the official docs
Two things to be clear about, and this page does not guess: first, the OpenAI Codex docs do not state availability by country or region, and regional access is out of scope here; second, the claim that an API key gives you every Codex cloud feature contradicts the docs: cloud requires ChatGPT sign-in, and features relying on workspace access or cloud services are limited or unavailable.
Subscription sign-in vs API key sign-in
ChatGPT subscription sign-in
Billed through your ChatGPT plan: the official pricing page lists Plus at $20/month and Pro plans at $100, $200 or $500 per month; usage consumes included plan credits, Pro plans currently have no five-hour limit, and Plus/Pro users who reach their limit can buy additional credits without upgrading.
API key sign-in
Billed by usage: OpenAI bills API key usage through your OpenAI Platform account at standard API rates; it covers local workflows in the desktop app, CLI and IDE extension, which the docs recommend for programmatic work such as CI/CD; features relying on ChatGPT workspace access or cloud services are limited or unavailable. The two differ in coverage, not in saving — this page does not compare which costs less.
Switching to API key sign-in (official steps)
First, the default CLI path is to run codex login and complete the browser flow. Second, to use an API key instead, pipe the key to codex login through stdin: printenv OPENAI_API_KEY | codex login --with-api-key. Third, in the IDE extension, on the signed-out screen select Use API Key, enter your key, then select OK. Fourth, run codex login status to see the active authentication method. Fifth, Codex caches login details locally in a plaintext file at ~/.codex/auth.json or in your OS-specific credential store, shared by the CLI and the extension. And keep the boundary in mind: all of this applies to local work only — Codex cloud requires signing in with ChatGPT.
Using it on QCode
On QCode, Codex goes through the API key path: Codex uses OpenAI Responses, with base_url = "https://api.qcode.cc/openai" + wire_api = "responses" in config.toml (as to API protocol type, Codex officially supports responses only for now). Through QCode.cc the Codex/GPT models are gpt-6.1-sol, gpt-6-sol, gpt-6-astra and gpt-5.6-terra. Claude Code and Codex share one QCode API key, and switching models needs no extra setup. Billed per token; per-model rates are listed on /models. The key starts with cr_ and goes into ~/.codex/auth.json.
Frequently asked questions
How do I use Codex without a ChatGPT subscription?
Sign in with an API key. The official docs state Codex supports two sign-in methods: Sign in with ChatGPT for subscription access, and Sign in with an API key for usage-based access. The ChatGPT desktop app, Codex CLI and IDE extension support both for local work; only Codex cloud requires signing in with ChatGPT, and an API key does not replace that.
How do I switch from subscription sign-in to an API key?
In the CLI, pipe the key to codex login through stdin: printenv OPENAI_API_KEY | codex login --with-api-key. In the IDE extension, on the signed-out screen select Use API Key, enter your key, then select OK. Run codex login status to see the active authentication method; login details are cached in a plaintext file at ~/.codex/auth.json or in your OS-specific credential store.
How is API key sign-in billed?
OpenAI bills API key usage through your OpenAI Platform account at standard API rates — with an API key, Codex uses standard API pricing instead of included ChatGPT plan credits. On QCode, the API key path is billed per token; per-model rates are listed on /models. These are two separate bills, and this page does not compare which saves more.
What do the ChatGPT subscription tiers cost?
The official pricing page lists Plus at $20/month and Pro plans at $100, $200 or $500 per month; Pro plans currently have no five-hour limit. Plus and Pro users who reach their usage limit can purchase additional credits to keep working without upgrading. These are the numbers published on OpenAI's own page, and the official page is authoritative.
What if I do not want a monthly plan?
QCode's Pay-as-you-go Packs: no monthly plan, from $4; top-ups start at $1. A pack is a one-time total quota valid for 7–90 days, used until it is gone, with no daily reset like Personal Plans; leftover quota is void at expiry, and buying any pack later revives the same key with no config change; concurrency is 5, the usable model list matches Personal Plans; billed per token; per-model rates are listed on /models.
How do I pay and get started from China or Russia?
QCode.cc is prepaid: you top up first, then buy a plan or a pack, and the purchase comes out of the balance; top-ups start at $1. Payment options include Alipay, WeChat Pay, Visa / Mastercard / AMEX, Apple Pay / Google Pay, and cryptocurrency. For Russia: SBP, a Russian bank card or other Russian local methods — you pay in rubles and the balance is credited in USD; these appear only when the interface language is Russian.
Sources
OpenAI's official Codex docs: the authentication page and the pricing page (learn.chatgpt.com, captured and checked on 2026-10-16). docs.qcode.cc: the Codex tutorial (updated 2026-10-15) and the Pay-as-you-go Packs, choose-a-plan and payment-methods pages (updated 2026-10-16). The qcode.cc Pay-as-you-go Packs page (all five languages, captured 2026-10-16).
Run Codex without a ChatGPT subscription
One QCode API key (starting with cr_) plus Codex API key sign-in: call gpt-6.1-sol, gpt-6-sol, gpt-6-astra and gpt-5.6-terra; billed per token; per-model rates are listed on /models.
Related reading
Claude Code Pay-as-you-go Packs: No Subscription, 7–90 Days
A one-time total quota valid for 7–90 days; extra packs add to the same key, and leftover balance can buy more packs or plans.
Using Codex in China: setup and payment (Oct 2026)
Your own provider in config.toml, base_url https://api.qcode.cc/openai when pointing at QCode, and top-ups via Alipay or WeChat Pay.
Codex 5-hour rolling limit is back (2026-07-30): mechanics and options
How the five-hour window is counted and four ways to keep going after hitting the limit.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
The information on this page was checked on 2026-10-16 against OpenAI's official docs and the docs.qcode.cc and qcode.cc pages; changes on the official side may happen without notice. Prices and limits on the OpenAI pricing page are the vendor's own published wording. QCode model availability is whatever /models shows.