New model · OpenAI Decisions API · public beta
GPT-6 Luna on Jev AI
GPT-6 Luna is the model behind OpenAI’s Decisions API. It reads text, JSON and images and answers typed questions: a probability for yes/no, one option with a probability for each, or a score on your rubric. On Jev AI it takes the same request as Jev, so a playground run, saved judge or API call switches to Luna by changing one field.
Checking connection…
What GPT-6 Luna is
OpenAI opened the Decisions API (POST /v1/decisions) to all API accounts in public beta on October 6, 2026, with gpt-6-luna as its only model. Instead of generating text it evaluates a list of questions against shared input and returns typed answers, which OpenAI describes as about ten times faster than the Responses API. Its three question types match Jev’s: predicate is Jev’s noul, and choice and score work the same way.
On Jev AI it runs alongside Jev, Clef, Laya and Mercury Decide. Jev AI sends every request straight to OpenAI and never answers a Luna request with another model. If OpenAI declines a question, the whole request fails before any charge instead of returning a partial answer.
How closely it matches Jev
We replayed every recorded text request on this site — 41 requests from the homepage demos and use-case templates — on GPT-6 Luna and compared the answers with the recorded jev-1.13.0 results. This measures agreement, not accuracy: the examples have no ground-truth labels.
Measured 2026-10-07, one request at a time from one client; median 0.52s and p90 0.73s wall-clock including the network. On the demos Luna counted 3,949 input tokens where Jev counted 3,749.
Use GPT-6 Luna with your Jev API key
Create a key on the Jev API page and store it as JEV_AI_API_KEY on your server. Use the Jev endpoint with gpt-6-luna, or the Decisions-compatible endpoint with openai/gpt-6-luna. Images are embedded PNG, JPEG or WebP data URLs, up to 4 per request.
curl --fail-with-body https://jev-ai.pro/api/v1/systemone \
-H "Authorization: Bearer $JEV_AI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6-luna",
"state": "Customer: I was charged twice for order 4821 and want my money back today.",
"questions": {
"team": {
"type": "choice",
"instructions": "Which team should handle this message?",
"criteria": {
"billing": "Payments, charges and refunds",
"shipping": "Delivery problems",
"technical": null
}
},
"urgent": {
"type": "noul",
"instructions": "Does this need a reply within the hour?"
}
}
}'Read answers.<id>.noul for the probability of yes, .choice for the selected option and .score for the rubric score, exactly as with Jev.
Limits and billing
- GPT-6 Luna uses 3× the Jev model rate: 3 credits per playground run, or 3 balance tokens per input token on the API. OpenAI charges $0.10 per million input tokens for it, more than twice Jev’s provider cost. Output tokens are free.
- Text, JSON and up to 4 images (4 MiB each, 8 MiB total). Each image adds input tokens.
- The Decisions API is in public beta at OpenAI. If it is unavailable or declines a question, the request returns an error before any charge.
OpenAI, GPT and GPT-6 Luna are trademarks of OpenAI. Jev AI is not affiliated with or endorsed by OpenAI. Details about the OpenAI product are from its public documentation as of 2026-10-07 and can change.