Skip to content

OpenAI cuts its API usage tiers from five to three

Build, Launch and Grow replace the old five-tier ladder, and the threshold for the top standard tier drops from $1,000 to $500 in total spend.

Source: OpenAI

OpenAI collapsed its five paid API usage tiers into three on 6 October: Build, Launch and Grow. Organisations move up automatically as total credit purchases cross each threshold, with no application or code change needed.

Build starts at $5 in total credit purchases, capped at $500 of usage a month. Launch needs $100 in purchases for a $5,000 monthly cap. Grow, the new top standard tier, now needs $500 in total payments rather than the old $1,000 — and raises rate limits to as much as 30,000 requests per minute and 180 million tokens per minute, depending on the model. A free tier still exists in allowed geographies, capped at $100 a month.

Why it matters: rate limits are what actually stop a side project from handling real traffic on the OpenAI API, and reaching the top tier now costs half as much cumulative spend as it did last week. Anyone who was tier-gated and had written code to request a limit increase can likely delete it — the new tiers apply automatically.

The caveat: exact RPM and TPM figures vary by model; check your account’s limits page rather than assuming the headline numbers apply to the model you use.

Share

Related reading

ML for Web Devs

Tokens, context windows, and why your bill is what it is

The unit of billing for language models is not the request or the word. Once you understand tokens and how a context window fills up, the invoice stops being a surprise.

ML for Web Devs

RAG is just search with extra steps

Retrieval-augmented generation gets written up like an architecture. It is a search query, a string concatenation and one API call, and the search query is the part that decides whether it works.

ML for Web Devs

A prompt is a spec, so write it like one

Prompt engineering has very little to do with magic words. It is the same skill as writing a clear ticket for a capable contractor who will not ask you any questions.

More from The Wire

All briefs

Anthropic cuts distillation attacks on Claude by up to 99%

New safeguards, including a reasoning classifier and stricter account limits, now block most attempts to copy Claude by training on its output. Blocked requests in high-confidence cases are billed, and some integrations may need updating.

via Anthropic