OpenAI API Usage Tiers: Build, Launch and Grow
As of 2026-10-08, the OpenAI API has three paid usage tiers, Build, Launch and Grow, which require $5, $100 and $500 in total credit purchases and carry monthly usage limits of $500, $5,000 and $200,000. The official changelog says they replaced five tiers on 2026-10-06. This page lists each tier's official rate limits, a before-and-after comparison with the old Tier 1–5, and where to check which tier you are on now.
Updated 2026-10-08
The new tiers in four numbers
Total credit purchases (Build / Launch / Grow)
The official rate-limits page words it as total credit purchases; once an organization reaches a threshold, its tier upgrades automatically.
Monthly usage limit
This is the monthly usage limit OpenAI approves for each organization, and the docs say it is separate from the spend limits you set yourself. The Free tier is $100 / month and requires an allowed geography, the same as on the official page archived on 2026-10-01.
TPM for Astra, Sol and Terra
RPM is 5,000 / 10,000 / 15,000 across Build, Launch and Grow; the Luna row reads RPM 5,000 / 10,000 / 30,000 and TPM 2M / 10M / 180M. The docs label these Standard rate limits, which differ from Ultrafast mode limits.
Top-tier threshold
According to the official rate-limits page as archived on 2026-10-01, the old top tier, Tier 5, required $1,000 paid; on the current official page the new top tier, Grow, requires $500 in total credit purchases. Both carry a $200,000 monthly usage limit.
What changed
As of 2026-10-08, the OpenAI API's paid usage tiers are Build, Launch and Grow: Build needs $5 in total credit purchases with a $500 monthly usage limit, Launch needs $100 with $5,000, and Grow needs $500 with $200,000. The OpenAI API changelog entry for 2026-10-06 reads "Simplified API usage tiers from five to three: Build, Launch, and Grow." and says organizations upgrade automatically as total credit purchases reach each tier's minimum. The Free tier matches the official page as archived on 2026-10-01: the user must be in an allowed geography, with a $100 monthly limit. The rate-limits page sums it up as higher tiers generally providing higher rate limits across models.
The official wording (2026-10-06 changelog)
The changelog line reads: "Organizations automatically upgrade as total credit purchases reach tier minimums." The rewritten rate-limits page says: "Your organization's usage tier upgrades automatically as its total credit purchases reach each threshold." The 2026-10-01 archive of the same page said instead: "As your spend on our API goes up, we automatically graduate you to the next usage tier.", and the qualification column moved from wording like "$5 paid" to "$5 in total credit purchases". The docs also point out that the monthly usage limit OpenAI sets for each organization is a separate control from the spend limits you configure on an organization or project.
Usage-tier change timeline
According to the official rate-limits page as archived on 2026-10-01, it still listed Tier 1 to Tier 5: thresholds of $5, $50, $100, $250 and $1,000 paid, with monthly limits of $100, $500, $1,000, $5,000 and $200,000.
The OpenAI API changelog announces that usage tiers are simplified from five to three, Build, Launch and Grow, with automatic upgrades by total credit purchases.
This page was checked against the official rate-limits page, the gpt-6-sol and gpt-6-luna model pages and the changelog; as of this date, Grow requires $500 in total credit purchases.
Confirmed vs not stated by OpenAI
Confirmed (verbatim in the official pages)
All of the following can be checked word for word on OpenAI's official pages: the simplification from five tiers to three on 2026-10-06; the three paid tiers are named Build, Launch and Grow; thresholds of $5 / $100 / $500 in total credit purchases; monthly usage limits of $500 / $5,000 / $200,000; the Free tier at $100 / month; RPM and TPM for the Astra, Sol, Terra row and the Luna row; rate limits apply at the organization and project level, not per user; Settings > Organization > Limits shows each model's limits at your current tier, and Upgrade tier in the Usage Tiers section moves you up. The old Tier 1–5 figures come from the 2026-10-01 archive of the same official page and the 2026-10-05 archive of the gpt-6-sol model page.
Unverified or not published
Three things are not stated on the official pages, so don't plan around them: first, which new tier each old Tier 1–5 account lands in — neither the docs nor the changelog give a mapping table, only the rule that tiers upgrade automatically by total credit purchases; second, the old column said paid and the new one says total credit purchases, and OpenAI does not explain whether the two are counted identically; third, limits under the new tiers for models other than Astra, Sol, Terra and Luna (image, audio, embeddings and so on) were not checked one by one here — your organization's Limits page is the reference.
Old tiers vs new, and the different kinds of limit
vs the old Tier 1–5
Comparing the official rate-limits page archived on 2026-10-01 and the gpt-6-sol model page archived on 2026-10-05 with the current official pages (a comparison of figures only; OpenAI has not published a mapping table): for the same $5, the monthly usage limit goes from $100 on Tier 1 to $500 on Build, and gpt-6-sol limits go from 500 RPM / 500,000 TPM to 5,000 RPM / 1,000,000 TPM, the same as the old Tier 2. At $100 in total, the old Tier 3 gave $1,000 / month and 5,000 RPM / 2,000,000 TPM, while Launch gives $5,000 / month and 10,000 RPM / 4,000,000 TPM, the same as the old Tier 4. At the top, Grow's gpt-6-sol limits of 15,000 RPM / 40,000,000 TPM match the old Tier 5; only the threshold dropped, from $1,000 to $500.
vs the spend limits you set
Your tier sets the OpenAI-approved monthly usage limit and the rate limits; a spend limit is a cost control you put on an organization or project yourself. Per the docs, a spend alert only sends a notification and traffic continues; once a hard spend limit is reached, affected requests return 429 with the code organization_spend_limit_exceeded or project_spend_limit_exceeded. For organization_usage_limit_exceeded you request a higher approved usage limit; credit_balance_exhausted means your prepaid credits ran out and you need to add credits.
What it means for developers: four checks
First, check your tier: Settings > Organization > Limits shows each model's rate limits at your current tier, and to move up you select Upgrade tier in the Usage Tiers section. Second, tell the errors apart: rate limits apply per organization and project, not per user, so read error.code first to see whether you hit a rate limit, the monthly usage limit, a spend limit or an empty credit balance. Third, ramp up even after an upgrade: the docs advise that once traffic reaches 1 million input tokens per minute, you grow it by no more than 50% every 15 minutes, and a slow_down error can arrive even within your RPM and TPM limits. Fourth, watch the response headers: remaining requests and tokens come back in HTTP headers such as x-ratelimit-remaining-requests and x-ratelimit-remaining-tokens; long-context models like GPT-5.5 have a separate rate limit for long-context requests.
Calling GPT models on QCode
If you want one key for GPT models, gpt-6.1-sol, gpt-6-sol, gpt-6-astra and gpt-5.6-terra are callable on QCode. Following the docs, set base_url in the OpenAI Python / JS SDK to https://api.qcode.cc/openai/v1; GPT models use the OpenAI-protocol endpoints only, not the Anthropic endpoint. Billing is per token, with each model's price on /models. QCode is prepaid: you top up first, then buy a plan, which is deducted from your balance; USD top-ups range from $1 to $2,000, payable by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay, WeChat Pay, Alipay or cryptocurrency. This page makes no promise about rate limits on the QCode side.
Frequently asked questions
How many OpenAI API usage tiers are there now?
Three paid tiers, Build, Launch and Grow, plus a Free tier. The OpenAI API changelog says on 2026-10-06 that five tiers were simplified to three; as of 2026-10-08 the thresholds are $5, $100 and $500 in total credit purchases, with monthly usage limits of $500, $5,000 and $200,000. The Free tier requires an allowed geography and is capped at $100 / month.
How much do I need to buy to reach Grow?
$500 in total credit purchases. Once you reach it, your organization moves to Grow automatically, with a $200,000 monthly usage limit. According to the official rate-limits page as archived on 2026-10-01, the old top tier, Tier 5, required $1,000 paid, with the same $200,000 monthly limit.
Which new tier do old Tier 1–5 accounts get?
OpenAI's docs do not publish a mapping from old tiers to new ones. The only rule you can confirm is that an organization's tier upgrades automatically by total credit purchases: $5, $100 and $500 correspond to Build, Launch and Grow. For your current tier and per-model limits, go by Settings > Organization > Limits.
What are the Build tier rate limits?
The official rate-limits page lists these Standard limits for Build: 5,000 RPM and 1,000,000 TPM for Astra, Sol and Terra, and 5,000 RPM and 2,000,000 TPM for Luna. Launch gets 10,000 RPM / 4,000,000 TPM and 10,000 RPM / 10,000,000 TPM respectively; Grow gets 15,000 RPM / 40,000,000 TPM and 30,000 RPM / 180,000,000 TPM.
Does a 429 mean my tier is too low?
Not necessarily. The docs describe at least three sources of 429: exceeding a rate limit; a slow_down error when traffic rises too fast, which can happen even within your RPM and TPM limits; and reaching a hard spend limit you set yourself (code organization_spend_limit_exceeded or project_spend_limit_exceeded). Read error.code first, then decide whether to back off and retry, raise your spend limit or aim for a higher tier.
Can I call GPT models on QCode with the same key?
Yes. gpt-6.1-sol, gpt-6-sol, gpt-6-astra and gpt-5.6-terra are callable on QCode with base_url set to https://api.qcode.cc/openai/v1, billed per token, with each model's price on /models. QCode is prepaid, and USD top-ups range from $1 to $2,000. This page makes no promise about rate limits on the QCode side; go by the actual responses you get.
Sources
Tier definitions, thresholds, monthly usage limits and the Standard rate table: the Rate limits page in OpenAI's developer docs (crawled 2026-10-08). Change date and wording: the 2026-10-06 entry in the OpenAI API changelog (crawled the same day). Per-model limits: the gpt-6-sol and gpt-6-luna model pages (same day). Old Tier 1–5 thresholds, limits and gpt-6-sol figures: the Internet Archive copies of the official Rate limits page (2026-10-01) and the gpt-6-sol model page (2026-10-05). Usage limits vs spend limits and the error codes: OpenAI's Spend limits page (crawled 2026-10-08). The QCode part: the billing, payment-methods and endpoints pages on docs.qcode.cc (crawled 2026-10-08).
One key for GPT models
gpt-6.1-sol, gpt-6-sol and other GPT models via https://api.qcode.cc/openai/v1, billed per token, prices on /models.
Related reading
OpenAI API outage, 503 and timeout guide
Read the status page first, then decide whether to change how you connect.
GPT-6.1 Sol API guide
Pricing, the 272K rule and how it differs from gpt-6-sol.
Pay for Claude API with Apple Pay / Google Pay
Top up your balance with a card already saved in your wallet.
The figures and quotes on this page were checked on 2026-10-08 against OpenAI's official changelog, rate-limits page and model pages; upstream changes may happen without notice. The old Tier 1–5 figures come from archives of the official pages dated 2026-10-01 and 2026-10-05. Your organization's actual tier and limits are whatever its Limits page shows; model availability on QCode is whatever /models shows.