- A separate account, ready to use: ready-made Pro 200 account ¥1,349 (≈ 201.05 USDT), Pro 200 already active. It’s hand-delivered: enter and verify a delivery email after paying and our team emails the account; the order page shows 30 minutes to about 4 hours. You can bind your own email.
- On your own account (Free, Plus or Pro; new subscription, renewal or upgrade): Pro 200 top-up ¥1,300, usually 5–30 minutes, same account and history.
- Need Ultrafast or the highest allowance: Pro 500 ¥3,599.
The allowances in Pro 200
| Pool | New subscription | Old allowance (until 29 Oct) | When it runs out |
|---|---|---|---|
| Chat: GPT-6 Pro | about 100 a week (Engadget) | 200 a week | Wait for the weekly reset; credits don’t cover chat |
| Chat: GPT-5.6 Sol Pro | not published | 170 a day (at most 200 a day with GPT-6 Pro) | Resets the next day |
| Codex / ChatGPT Work (all models) | about 10× Plus (Engadget) | about 20× Plus | Wait for the weekly reset, or use credits |
Pro plans currently have no five-hour limit in Codex; the weekly allowance is what runs out. In September OpenAI published per-5-hour estimates for the old 20x — GPT-6 Sol 300–3,000, Luna 7,000–56,000, Astra 100–900 local messages — which are still useful as proportions: the smaller the model, the further the allowance goes. The pools don’t eat into each other, and that is what makes planning possible.
Which allowance for which job
| Task | Where and which model | Why |
|---|---|---|
| Hard reasoning, long-document analysis, work you deliver | GPT-6 Pro in chat | Only about 100 a week on a new subscription (reported) — keep them for what’s worth it |
| Heavy everyday questions, writing, editing | GPT-5.6 Sol Pro in chat | Spares the weekly GPT-6 Pro allowance (170 a day on the old allowance) |
| Mainline coding and features | Codex: GPT-6 Sol | The best value: about 1/5 of Astra’s credit cost for the same output |
| Bulk renames, tests, docs, well-defined small edits | Codex: GPT-6 Luna | About 1/20 of Sol’s cost; hard to exhaust |
| The hardest debugging, long-horizon agent work | Codex: GPT-6 Astra | Strongest and priciest: about 5× Sol for the same output |
The ratios come from OpenAI’s credit rate card: per million output tokens, Astra 1,250, Sol 250 and Luna 12.5 credits — Astra about 5× Sol, Sol about 20× Luna. OpenAI’s September per-5-hour estimates pointed the same way.
Your first week
| When | What to do | Why |
|---|---|---|
| Day 1 | Sign in at chatgpt.com and confirm GPT-6 Pro is in the model menu; sign in to Codex with the ChatGPT account (CLI, IDE extension, desktop app or web). On a ready-made account, change the password first; you can also bind your own email. | Confirms your allowances and that only you can get in |
| Day 2 | Write custom instructions; if you need old chats, export conversations.json from the old account and upload it to a chat in the new one as reference. | Less allowance spent re-explaining yourself |
| Day 3 | Add AGENTS.md at the repo root: conventions, common commands, folders not to touch. | Codex reads it every run, so you don’t repeat it per task |
| Days 4–5 | Assign models as in the table above: Sol for mainline work, Luna for bulk, Astra only for the hardest; take deep questions to GPT-6 Pro in chat. | The same allowance does several times the work |
| Days 6–7 | Note which cap you hit: GPT-6 Pro in chat or the weekly Codex allowance. Rebalance; add credits if Codex occasionally runs short. | Know the bottleneck before you max it out |
What moves between accounts and what doesn’t: switching to a ready-made account.
7 habits that waste Pro 200
- GPT-6 Pro for everything. About 100 a week on a new subscription (reported) disappears by midweek if you spend them on trivia. Use GPT-5.6 Sol Pro for everyday questions.
- Astra for simple edits. The same output costs about 5× more than on Sol — pointless for renaming a variable.
- Fast mode by default. Under OpenAI’s billing rules Fast mode uses your included allowance 2.5× as fast as Standard (2× once you are on credits). Turn it on only when you’re in a hurry.
- Re-pasting whole files. Cached input counts at 10% of normal input; continuing in the same session beats starting over and pasting again.
- No AGENTS.md. Re-explaining the project in every task spends allowance on repetition.
- Sharing the account. Ready-made accounts are one person per account; sharing breaks OpenAI’s terms and invites risk checks.
- Letting it lapse. Each ready-made order is 30 days, with reminders 7, 3 and 1 days before expiry; renew by messaging Telegram @ccnb1937 with your order ID and keep the same account. If you are on the old allowance, a lapse before 29 October loses it: a new subscription gets the new allowance.
Don’t have Pro 200 yet?
From 10 to 30 September 2026 OpenAI paused new sign-ups to the $200 tier, and a ready-made account was the only way in; since 30 September anyone can subscribe again, with the new allowance. Three routes:
- A separate account, ready to go: a ready-made Pro 200 account (¥1,349 ≈ 201.05 USDT) with Pro 200 already active. It’s hand-delivered: after paying, enter a delivery email on the order page and click the verification link; our team emails the account, and the order page shows 30 minutes to about 4 hours. One person per account; you can bind your own email; renew every 30 days.
- Keep your own account (Free, Plus or Pro): Pro 200 on your own account (¥1,300), new subscription, renewal or upgrade, usually 5–30 minutes after the checks on the order page.
- Pro 200 still active from before 30 September: keep it renewing without a gap; that holds the old allowance through 29 October.
The pricing page also lists a limited ready-made account still on the previous 20x allowance; per OpenAI that allowance lasts through 29 October 2026 while the subscription stays active.
Contact the bottom-right live chat promptly if anything goes wrong. Need Ultrafast or the highest cap: Pro 500. If you mainly want more Codex and don’t need GPT-6 Pro in chat, see alternatives to 20x.