free
Free allowance
Every early Decisions API key starts with free calls, enough to measure accuracy on your own data.
Decisions API pricing
Usage-based and billed on input tokens, with a free allowance on every early key. Here is how a Decisions API call is priced, and how the published rates in this category compare.
Plans
How the Decisions API is billed
Exact rates go to the waitlist before the first keys are issued. The structure is fixed:
pay as you go
Per input token
You pay for the context you send. The answer is a few fixed fields, and failed requests are free.
volume
Committed volume
Running millions of decisions a month? Tell us on the waitlist and we will agree rates up front.
Cost per call
What a Decisions API call costs
Because the Decisions API bills on input, cost per call is set almost entirely by how much context you send. Typical sizes:
| Context | Input tokens per call |
|---|---|
| A short support ticket | ~80 |
| A customer email | ~400 |
| A product listing with attributes | ~700 |
| A 20-message chat transcript | ~2,000 |
| An agent’s full working state | ~8,000 |
Asking several questions about the same context in one request costs the same input once, not once per question. That is usually the cheapest way to use the Decisions API.
Published rates · updated September 30, 2026
OpenAI Decisions API pricing and Jev pricing
What each company has published so far.
| OpenAI Decisions API | TypeSafe Jev | |
|---|---|---|
| Input | Not announced | $0.042 / 1M tokens |
| Output | Not announced | Free |
| What coverage assumes | Standard GPT-6 Luna rates (reported, unconfirmed) | — |
| Availability | Limited preview | Early access |
Sources: OpenAI DevDay 2026 recap, DEV Community roundup (reported Luna rates), Every (pricing not announced), TypeSafe blog.
Calculator
Estimate your monthly Decisions API cost
Input cost only, at published or reported per-token rates.
TypeSafe Jev $0.042 / 1M, published
—
OpenAI, if standard Luna input rates apply $0.10 / 1M, unconfirmed
—
decisionsapi.cc rates sent to the waitlist
FAQ
Decisions API pricing FAQ
Is there a free tier for the Decisions API?
Yes. Every early Decisions API key starts with a free allowance, enough to wire up a real workflow and measure accuracy on your own data before you pay anything.
Are output tokens billed?
Decisions API billing is based on input tokens. The answer is a few fixed-shape fields, so output adds almost nothing to a call.
Do failed requests count?
No. Requests that return a 4xx or 5xx error are not billed.
How much does the OpenAI Decisions API cost?
As of September 30, 2026, OpenAI has not published separate pricing for its Decisions API, which is in limited preview. Some coverage assumes standard GPT-6 Luna token rates apply; OpenAI has not confirmed this.
How much does Jev cost?
TypeSafe lists Jev at $0.042 per million input tokens, with output tokens free.
Do you offer volume pricing?
Yes. If you expect to run millions of Decisions API calls a month, say so on the waitlist and we will talk rates before your key is issued.