OpenAI introduces a faster decision API and a new Google Cloud access route

The Decisions API is generally available, and OpenAI has announced model and Codex distribution through Google Cloud.

OpenAI's announcements this week address two different deployment questions: how quickly an application can obtain a bounded decision, and how an organisation can purchase and govern access through its cloud environment.

A specialised path for bounded questions

The Decisions API became generally available on 22 October. Built on GPT-6 Luna, it accepts text, images and structured input for questions with a bounded answer space. OpenAI says it can respond up to ten times faster than using GPT-6 Luna through the Responses API for these tasks.

The release includes native JSON input, prompt caching, images in instructions and batches of up to 400 questions per request. It also adds native integration with GPT-Live and US and EU inference residency. The examples span routing, classification and in-conversation decisions where waiting for a longer general-purpose response can interrupt the surrounding interaction.

One customer example comes from Speak, which reports reducing time to the next action from 1.4 seconds to 0.13 seconds using the GPT-Live integration. That illustrates the product's intended role: deciding what happens next inside a broader application, rather than generating every part of the user experience through the same request.

Procurement through an existing cloud relationship

On 21 October, OpenAI announced that its frontier models and Codex workflows are coming to Google Cloud's Model Garden. The planned integration uses Google Cloud access controls, data-governance arrangements and infrastructure. Organisations will be able to buy through Google Cloud Marketplace and use eligible cloud commitments.

General availability is planned in the coming weeks. The announcement also describes future model choice in Google's Antigravity and Gemini Agent products. Microsoft remains OpenAI's primary cloud partner.

For enterprise teams, these announcements affect separate layers of a deployment. The application can choose a specialised decision endpoint for a latency-sensitive step, while procurement and platform teams choose the commercial and governance route through which access is managed. Keeping those decisions explicit helps teams explain why a particular model is called, where it runs and how the service fits their existing cloud estate.

The resulting pattern is more modular: a fast decision component inside the product, a general reasoning workflow where it is needed and an enterprise access layer aligned with the customer's cloud operations.

Sources

- [Build more responsive apps with the Decisions API](https://openai.com/index/decisions-api/) — OpenAI · 2026-10-22 - [OpenAI models are coming to Google Cloud](https://openai.com/index/openai-models-google-cloud/) — OpenAI · 2026-10-21