OpenAI releases Decisions API, Jev competitor, now generally available with voice and visual modes at $0.10 per million input tokens and no output fees (Tim Fernholz/TechCrunch)
openai voice
| Source: Techmeme | Original article
OpenAI has made its Decisions API, a Jev competitor, generally available, adding voice and visual modalities and pricing it at $0.10 per million input tokens with no output‑token fees.
OpenAI has moved its Decisions API out of beta and made it generally available, adding voice and visual input capabilities while pricing the service at $0.10 per 1 million input tokens with no charge for output tokens. The model, built on the compact GPT‑6 Luna engine, promises decision‑making speeds up to ten times faster than the standard Responses API and returns structured answers such as probability predicates and ranked choices.
The launch positions OpenAI directly against TypeSafe’s Jev and the open‑source Laya model, extending the nascent “decision‑only” segment that OpenAI first unveiled at DevDay 2026. By supporting text, image and now voice inputs, the Decisions API broadens the toolkit for developers who need rapid, low‑cost inference for tasks like recommendation ranking, risk assessment or real‑time control loops. The pricing structure—charging only for input tokens—further lowers the barrier for high‑volume applications, potentially accelerating adoption in sectors ranging from fintech to IoT.
As we reported on 23 October 2026, OpenAI’s entry into the decision‑model market intensified competition among fast, cheap intelligence providers. The new modalities and transparent pricing deepen that rivalry and may force rivals to adjust their own cost and latency targets.
What to watch next includes early performance benchmarks from developers integrating the API, any refinements to the pricing model, and how competitors such as Jev respond with feature or price updates. OpenAI’s roadmap also hints at a full GA rollout in the coming weeks, so the next wave of announcements will likely focus on broader SDK support, enterprise SLAs and possible extensions into safety‑critical use cases.
Sources
Back to AIPULSEN