OpenAI Slashes API Tiers From Five to Three, Cuts Top Threshold in Half
OpenAI collapses its five paid API usage tiers into three, cutting the spend needed to reach the top rate-limit tier in half.
- OpenAI replaces five paid API tiers (Tier 1-5) with three: Build, Launch, and Grow.
- Top tier Grow now unlocks at $500 cumulative API spend, down from $1,000 for the old Tier 5.
- Existing paid organizations are migrated automatically, with no action required from developers.
- New tier assignments visible on the organization limits page in the OpenAI dashboard.
- Change simplifies the rate-limit ladder that governs RPM, TPM, and batch queue caps per model.
- Move contrasts with OpenAI's rising consumer prices, including a new Pro 500 plan at $500 per month.
OpenAI cuts paid API usage tiers from five to three
OpenAI has consolidated its five paid API usage tiers into three levels named Build, Launch, and Grow, according to an official announcement. The company also reduced the cumulative payment threshold for its highest standard tier from $1,000 to $500.
OpenAI says existing paid organizations will move to the new structure automatically. The migration requires no application, support request, or code change. Updated assignments and model-specific limits appear on the dashboard’s limits page.
Five gates become three
The outgoing system assigned organizations to Tiers 1 through 5 using cumulative API payments and, for higher levels, the time since the first successful payment.
| Tier | Cumulative payments | Minimum payment history |
|---|---|---|
| Tier 1 | $5 | None |
| Tier 2 | $50 | 7 days |
| Tier 3 | $100 | 7 days |
| Tier 4 | $250 | 14 days |
| Tier 5 | $1,000 | 30 days |
The replacement structure groups paid organizations by stage:
- Build covers prototyping and early traffic.
- Launch covers applications entering production.
- Grow provides the highest standard usage limits.
The $500 Grow requirement refers to cumulative API payments by the organization. ChatGPT subscriptions and monthly spending commitments are separate. OpenAI’s announcement concerns usage limits and does not describe a change to per-token model prices.
Grow arrives at half the spend
Rate limits control how much traffic an organization can send within a given period. OpenAI measures capacity through limits such as requests per minute, tokens per minute, requests per day, and queued tokens for batch workloads. Exceeding an applicable limit produces an HTTP 429 response, sometimes with a Retry-After header.
Reducing Grow’s payment threshold by 50% lets organizations qualify for the highest standard tier earlier. That can provide more room for production launches, traffic spikes, parallel agent tasks, and large batch jobs.
Each model retains its own ceilings within a tier. A Grow organization may therefore have different request and token allowances for different models, while Batch API capacity can follow separate queue limits. The dashboard remains the authoritative source for an organization’s effective capacity.
Audit the limits that apply
Automatic migration makes the dashboard review especially important for teams planning launches or capacity-sensitive workloads. Developers should verify:
- Tier assignment: Confirm whether the organization moved to Build, Launch, or Grow.
- Per-model capacity: Compare RPM and TPM limits with peak traffic and token usage, including short bursts.
- Daily ceilings: Check request-per-day limits for sustained workloads.
- Batch capacity: Review queued-token limits separately from synchronous API limits.
- Project configuration: Confirm that production projects and API keys belong to the intended organization.
API clients should continue to handle 429 responses with bounded retries, exponential backoff, jitter, and support for Retry-After. Queues and concurrency controls remain necessary because a higher tier increases available capacity without eliminating model-specific limits or sudden demand spikes.
Comments
