OpenAI API Usage Tiers: Build, Launch, and Grow Limits

OpenAI Application Programming Interface (API) usage tiers set an organization's monthly usage limit and rate limits based on cumulative credit purchases rather than consumed spend. As of October 2026, the structure includes a Free tier for eligible regions alongside three paid tiers named Build, Launch, and Grow.

Each tier level raises the monthly usage limit and expands standard rate limits across models. Organizations qualify for higher tiers automatically when cumulative account payments meet defined financial thresholds.

On October 6, 2026, OpenAI restructured its paid tiers from five legacy brackets down to three consolidated tiers. This structural change cut the entry threshold for maximum standard API rate limits by 50 percent.

OpenAI API Usage Tiers and What Each Requires

OpenAI organizes API access into four active usage tiers that define an organization's monthly usage limit and rate limits according to historical credit purchases. The structure begins with a Free tier for accounts in supported regions, followed by the paid Build, Launch, and Grow tiers.

Tier progression depends on cumulative payments added to the account. Free tier access needs an allowed geography and carries a $100 monthly usage limit. The paid tiers need $5, $100 and $500 in total credit purchases, for monthly limits of $500, $5,000 and $200,000. The list below pairs each figure.

Monthly usage limits are approved per organization, as detailed in the official OpenAI rate limits documentation. These limits function independently from the user-configured spend limits that administrators set to control monthly budgets.

• Free tier: Requires account presence in an allowed geography, granting a $100 monthly usage limit.
• Build tier: Requires $5 in total credit purchases, granting a $500 monthly usage limit.
• Launch tier: Requires $100 in total credit purchases, granting a $5,000 monthly usage limit.
• Grow tier: Requires $500 in total credit purchases, granting a $200,000 monthly usage limit.

The October 2026 Tier Consolidation

OpenAI consolidated its five legacy paid usage tiers into three streamlined tiers on October 6, 2026. The three tiers are Build, Launch, and Grow, with Grow as the new highest tier.

In a public announcement on the OpenAI Developer Community, OpenAI noted that this consolidation makes it easier to qualify for higher OpenAI API rate limits. Qualifying for the top tier previously required $1,000 in account payments under the retired framework, whereas the Grow tier requires $500 in total API payments.

Organizations holding an active paid tier were transitioned to their corresponding new tier automatically after the October 6, 2026 announcement. The community notice confirmed that account administrators needed to take no manual action, though OpenAI did not publish the specific qualification thresholds or mappings for the retired intermediate tiers.

Per-Model Rate Limits for Astra, Sol, Terra, and Luna

OpenAI assigns requests per minute (RPM) and tokens per minute (TPM) caps to each tier based on the specific model requested. Higher tiers provide higher rate limits across standard models.

Standard rate limits differ across model families. Deployments targeting the Astra, Sol, and Terra models share identical throughput caps within each tier, while the Luna model receives higher token capacity. For what the models cost per token, see GPT-6 Astra pricing, GPT-6 Astra explained, GPT-5/6 Luna explained and GPT-5/6 API pricing.

The table gives standard rate limits. OpenAI lists Ultrafast rate limits separately, and those figures are not published here.

• Build tier (Astra, Sol, and Terra): 5,000 requests per minute (RPM) and 1,000,000 tokens per minute (TPM).
• Build tier (Luna): 5,000 RPM and 2,000,000 TPM.
• Launch tier (Astra, Sol, and Terra): 10,000 RPM and 4,000,000 TPM.
• Launch tier (Luna): 10,000 RPM and 10,000,000 TPM.
• Grow tier (Astra, Sol, and Terra): 15,000 RPM and 40,000,000 TPM.
• Grow tier (Luna): 30,000 RPM and 180,000,000 TPM.

Credit Purchases vs Spend and Account Upgrade Steps

Tier progression is determined by cumulative credit purchases rather than consumed token spend. An organization's usage tier upgrades automatically as its total credit purchases reach each threshold, though OpenAI does not publish how long an upgrade takes to register.

Some coverage uses the shorthand that a team must spend $500 to unlock maximum standard rate limits. That description is imprecise. The official OpenAI rate limits guide specifies total credit purchases as the governing metric, while the developer community update cites total API payments. Because qualification counts total credit purchases rather than spend, credit bought but not yet used still counts toward the threshold, though upgrade timing is not published.

You can inspect current limits under Settings. To upgrade, select Upgrade tier in the Usage Tiers section.

• Go to Settings.
• Select Organization from the navigation menu.
• Select Limits.
• Select Rate limits to view per-model limits.
• Select Upgrade tier in the Usage Tiers section.

Throughput Multipliers, Operational Fit, and Unpublished Limits

Advancing from the Build tier to the Grow tier delivers substantial throughput expansion across both request counts and token processing caps. For Astra, Sol, and Terra, tokens per minute expand from 1,000,000 to 40,000,000 (a 40-fold increase), while request allowances expand from 5,000 to 15,000 RPM (a 3-fold increase). For Luna, tokens per minute expand from 2,000,000 to 180,000,000 (a 90-fold increase), and request limits rise from 5,000 to 30,000 RPM (a 6-fold increase).

Organizations can reach the Grow tier's standard limits for $500 in total credit purchases, half the previous $1,000 top threshold.

Several operational metrics remain unpublished by OpenAI. The documentation does not state per-minute rate limits for the Free tier, nor does it state how many minutes or hours an automated upgrade takes to register after a payment clears. OpenAI also does not publish numerical batch queue limits per tier, noting only that batch queue limits are calculated based on the total number of input tokens queued for a given model, with tokens from pending batch jobs counting toward the queue limit until completion.

Check your current limits under Settings, Organization, Limits before buying more credit, and confirm figures on OpenAI's rate limits guide.