OpenAI's Codex pricing page includes a time-limited Codex usage promotion for Pro tiers. The UI must show the end date and avoid evergreen claims.
high impact · checked 2026-05-21 · confidence high
View sourceusage_page · codex usage
Codex usage explained
Codex usage can mean different things: local CLI messages, cloud tasks, ChatGPT plan access, credits, or API-token billing. This page explains those modes separately, shows what may count against a plan, and helps developers decide when to use subscription access or API fallback.
Independent guide. We are not affiliated with or endorsed by Codex usage explained. Prices and limits may change. Each claim below should link to a provider source and show a last-checked date.
Cost scenarioestimate, not billing quote
$20/mo estimate
Calculator results are estimates based on your inputs and public pricing assumptions. They are not billing quotes and may not match your actual provider invoice or account-specific limits.Plan fit: Pro. Limit risk: low. Intensity score: 29. Mode: subscription.
Open pricing guidePricing and limits table
Every row is rendered with provider names as plain text, source links, last-checked dates, confidence, and caveats. Prices and usage limits can change; verify official sources before purchasing.
| Tool | Plan | Monthly | Best fit | Source / freshness | Caveat |
|---|---|---|---|---|---|
| Codex | ChatGPT Free + Codex | $0/mo | trial, light tasks | Source Checked 2026-05-21 · high | Capabilities and trial limits can change; verify source before publishing exact limits. |
| Codex | ChatGPT Go + Codex | $8/mo | lightweight coding tasks | Source Checked 2026-05-21 · high | Go plan may vary by region and product availability. |
| Codex | ChatGPT Plus + Codex | $20/mo | solo developer, few focused coding sessions per week | Source Checked 2026-05-21 · high | Included limits vary by model/local/cloud/code-review mode; promo periods can change effective allowance. |
| Codex | ChatGPT Pro 5x + Codex | From $100/mo | longer higher-intensity sessions, Codex-heavy individual | Source Checked 2026-05-21 · high | Promo is date-bound; UI must render promo end date and avoid evergreen claims. |
| Codex | ChatGPT Pro 20x + Codex | $200/mo | very heavy Codex user, daily high-intensity agent sessions | Source Checked 2026-05-21 · high | Promo is date-bound; Plus baseline and exact limits depend on model and usage mode. |
| Codex | Codex with API key | Token/API based | CI automation, shared environments, programmatic workflows | Source Checked 2026-05-21 · high | API mode does not equal ChatGPT subscription mode and may delay access to some Codex models/features. |
Answer first: what this page helps you decide
Codex usage can mean different things: local CLI messages, cloud tasks, ChatGPT plan access, credits, or API-token billing. This page explains those modes separately, shows what may count against a plan, and helps developers decide when to use subscription access or API fallback.
Plan and usage table
Show plan rows with price_display, billing_period, usage_or_limit_notes, region/tax caveat, source_url, last_checked, confidence, and correction link.
Plan factsUsage notesSource and confidence
Real cost scenarios
Compare light solo developer, heavy agent user, team lead, and API-heavy builder scenarios. State assumptions and link to calculator result.
Light solo developerHeavy agent userTeam or reimbursement caseAPI fallback case
What may change
Explain that prices, usage pools, reset windows, model access, taxes, regional availability, and enterprise terms may change. Link to changelog and contact correction route.
Last checked policySubmit a correction
Cloud, local, and API usage are different budgets
Codex usage can create confusion because developers often mix several workflows in one mental budget. A cloud task, a local coding session, and an API-backed automation may all help write code, but they can be governed by different limits, product surfaces, billing meters, and organizational approval paths. Before choosing a plan, write down where the work will actually happen: inside a managed cloud environment, inside an editor or terminal, or inside a custom script that calls an API. Then estimate the number of sessions, the expected task length, and the failure rate that causes retries. A workflow with many small edits may feel inexpensive, while a workflow that repeatedly asks for broad repository analysis can create usage pressure faster than the headline plan price suggests.
Cloud tasksLocal CLI workAPI billing
When to recheck the source before relying on Codex limits
Recheck the official source when a team is about to standardize on Codex, when a reimbursement policy is being written, or when a developer reports throttling that the pricing page did not make obvious. Public plan descriptions can lag product experiments, regional availability, workspace controls, or preview features. Treat this page as an independent decision aid: it keeps source URLs, last-checked dates, confidence labels, and caveats visible, but it is not an official billing statement. If your usage is sensitive, run a small pilot, document the observed limits, and compare the result against Claude Code, Cursor, and Copilot before making a larger purchase.
Fast-changing policyAccount-specific rulesTeam rollout
Latest pricing and limit changes
Pricing changes are part of the product, not a footnote. Check the dated changelog before making a plan decision.
GitHub Docs says Copilot individual plans move to AI Credits on June 1, 2026. P1 Copilot pages must separate current premium requests from upcoming AI Credits.
high impact · checked 2026-05-21 · confidence high
View sourceClaude Help confirms Max 5x and 20x pricing and usage positioning for Claude Code users.
medium impact · checked 2026-05-21 · confidence high
View sourceIs this an independent OpenAI pricing guide?
Yes. This is an independent guide and is not affiliated with or endorsed by OpenAI. Brand names are used only to identify the product being discussed.
Can I rely on this as my final bill?
No. Calculator and table outputs are estimates based on public assumptions. Your actual bill or account limits may vary by region, plan, taxes, promotions, and provider policy changes.
What should I check before upgrading?
Check plan price, included usage or credits, reset behavior, model access, team controls, overage or on-demand terms, source link, and last-checked date.
Why do users hit usage limits even on paid plans?
Paid plans can still have capacity, credit, model, or reset-window constraints. Heavy agent sessions, long context, cloud tasks, or advanced models may consume more capacity.
Does this page collect my account usage?
No. P0 should not connect to provider accounts or collect tokens, billing credentials, API keys, or private usage dashboards.
How do I report outdated pricing?
Use the contact or correction link with a source URL. Updated facts should create or update a changelog entry.
Compare pricing, limits, and workflow fit
Calculator hubclaude code pricingclaude code usage limitscodex pricingcodex usagecursor pricingclaude code vs codexclaude code vs cursorpricing changelogclaude code costcodex limitscodex vs cursorapi vs subscriptiongithub copilot pricinggithub copilot vs cursorclaude max vs prochatgpt pro vs claude maxcodex pricing changelogclaude code pricing changelog