OpenAI's GPT-6 Models Arrive on Google Cloud in Private Preview
· Cortex Solutions
On October 21, Google Cloud began offering OpenAI's GPT-6 models in private preview, joining Gemini and Claude. What changes for Saudi businesses choosing AI models.
Google Cloud announced on October 21, 2026 that it has started offering OpenAI models to its customers through Model Garden. According to its official announcement, this makes it the only hyperscale cloud provider offering Gemini, Claude and GPT on one enterprise platform, alongside more than 200 first-party, third-party and open models.
The announcement was published on the Google Cloud blog by Michael Gerstenhaber, VP of Product Management for Gemini at Google Cloud, and was covered the same day by the AI news site The Decoder in a report by Matthias Bastian.
Which models are included?
According to Google Cloud, the first batch includes three OpenAI models:
- GPT-6 Astra: built for demanding work such as coding, computer use, professional tasks and scientific research.
- GPT-6.1 Sol: near-Astra intelligence at a lower cost, for agentic coding, computer use and professional work.
- GPT-6 Luna: a fast, cost-efficient model for everyday work at scale.
The announcement stresses that customers can select and swap models based on performance, latency and cost, "without architectural lock-in or separate vendor contracts."
Private preview, not general availability yet
The details matter: the models are available now in a private preview for select enterprise customers only, and interested companies can register through a form on the Google Cloud website. The announcement gives no date for general availability and does not name the regions where the models will run; it only says "more coming soon."
The announcement also says OpenAI models are coming "soon" to Gemini for business and to the Google Antigravity coding tool, without a timeline. What is available from the same day, according to Google, is the ability to connect OpenAI's Codex CLI and desktop workflows directly to the Gemini platform for inference.
Billing and contracts
The announcement puts strong emphasis on the commercial side. According to Google Cloud:
- OpenAI models are purchased directly through Google Cloud Marketplace.
- All usage appears on a single Google Cloud invoice.
- Usage draws down the customer's existing cloud commitments and counts toward Flexible Savings Plans across model providers.
- No separate vendor contract is required.
Detailed per-model pricing has not been published yet.
Data and privacy
Google Cloud says customer prompts, context and outputs "remain entirely within your designated data perimeter and are never used to train foundation models," a point also reported by The Decoder. The announcement adds that OpenAI models run on dedicated Google Cloud infrastructure, that queries pass through isolated gateways, and that every agent and model interaction runs under an authenticated identity with unified audit logging. However, it describes Zero Data Retention options and Confidential Compute as "coming soon," meaning they are not available yet.
Why this news matters
Gerstenhaber said in the announcement that "Google Cloud has always been defined by choice, enterprise-grade governance, and market-leading price-performance," and that the foundation of openness is giving customers the flexibility to run agentic workloads across their choice of models and infrastructure to balance cost, capability and performance.
The Decoder points out a notable twist: Google is offering its customers models from a direct competitor in the model market, next to the Gemini models it builds itself. The message is clear: the cloud market is moving toward platforms that are neutral about the model and compete on security, governance and price rather than locking customers into a single model.
The announcement also quotes Yashodha Bhavnani, VP of AI Products at Box, saying that using OpenAI models through Google Cloud will give their customers access to these models together with Google Cloud's security and compliance capabilities.
What does it mean for Saudi businesses?
- Choice is becoming the standard: a business no longer has to build its systems around one model. It is better to design solutions so the model can be swapped later based on cost and quality, without rebuilding everything.
- Ask where your data lives before asking about price: the announcement does not yet name regions, and some privacy guarantees are still "coming soon." Organizations subject to personal data protection requirements in the Kingdom should verify where processing happens and the data retention terms before any production use.
- A single invoice simplifies management: if your business already runs on Google Cloud, consolidating several models under one bill and contract reduces administrative and financial complexity.
- A preview is not a launch: do not build an operating plan on a private-preview service with no clear general availability date. Test, measure, then decide.
- The value is in the application, not the model: a model alone does not solve a problem. Real value comes when AI is connected to the organization's own data and processes: customer service, sales, accounting and operations.
At Cortex Solutions we build AI agents connected to your tools to automate customer service, document processing and workflows, with simple controls and clear reports on what they have done. Learn more: https://cortex.sa/services/ai-automation
Sources: Google Cloud's announcement on its official blog, October 21, 2026, and The Decoder's report, October 21, 2026.