Decision models · Reviewed September 30, 2026
Give software a bounded decision it can act on.
A routing task needs a category or a next step. Jev has a documented decision API; coverage of OpenAI’s September 29 DevDay describes a new Decisions API in limited preview. This comparison separates published documentation from launch reporting.
This is an independent comparison. This website is not affiliated with OpenAI or TypeSafe. OpenAI Decisions API is not integrated into this playground or API.
Evidence and open questions
The OpenAI changelog and endpoint index checked on September 30 did not list a dedicated Decisions API reference. Reported demo numbers do not establish a production SLA or a head-to-head benchmark.
Official input price
Jev
$0.042 / 1M
OpenAI Decisions API
Unconfirmed
TypeSafe lists $42 per billion input tokens, equivalent to $0.042 per million. This is the vendor’s price; this independent service uses its own credit rates. OpenAI preview pricing was not verified.
TypeSafe pricingLatency evidence
Jev
Test your cases
OpenAI Decisions API
~150 ms*
*Launch coverage reports an approximately 150 ms demonstration on a specialized GPT-6 Luna version. This is not an SLA. We have no matched workload, region, concurrency, or p95 comparison with Jev.
Launch report · secondary sourceProbability calibration
Jev
RLCD
OpenAI Decisions API
Unconfirmed
TypeSafe describes training for calibrated decisions with RLCD. No Decisions API-specific calibration method was found in the checked OpenAI docs. That does not prove its probabilities are uncalibrated.
TypeSafe explanation of RLCDA bounded decision
Define the available branches and the evidence that separates them. A decision model returns a signal your code can use. A chat model with structured outputs can also fit a schema, but a valid JSON shape alone does not establish decision quality or calibration.
BILLING_SUPPORTTECH_SUPPORTREFUNDGENERAL_SUPPORTChoose around your workload
01
Start with its published Noul, Choice, and Score contract. In the playground, test normal, ambiguous, and incomplete inputs against your own expected answers.
02
Launch coverage describes text and image context for OpenAI. Once official access and image limits are clear, test product categories, generated-image pass/retry decisions, or document routing with visual evidence. These are potential uses, not features offered here.
03
Compare error rates, probability calibration, cost, and p50/p95 latency with representative labeled data. Define a review threshold and a fallback in application code. A confident score does not guarantee a correct decision.
From evaluation to integration
Jev · TypeSafe System One model
TypeSafe documents POST https://api.typesafe.ai/v1/systemone. Requests carry state, model, and typed questions; responses return answers keyed by question ID.
OpenAI · reported preview
Pasquale Pillitteri’s launch report describes a GPT-6 Luna specialization that chooses from finite answers using text or image context. These details remain launch reporting in this comparison.
Side-by-side comparison
Sources and scope · September 30, 2026
OpenAI launch details above are attributed to Pasquale Pillitteri’s September 29 report. We checked OpenAI’s public changelog and endpoint index; neither provided a dedicated Decisions API entry at review time. Jev’s contract, confidence behavior, and price come from TypeSafe’s own material. No side-by-side test was run. Check each provider’s latest documentation for current availability and prices.