The OpenAI Google Cloud launch landed on October 21, 2026. GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna entered private preview in Google’s Model Garden for select enterprise customers, with general availability promised “in the coming weeks.” Codex CLI and Desktop can already route inference to Google. Neither company disclosed a price. On OpenAI’s own API, the three models run from $0.10 to $10 per million input tokens.
Key takeaways
- Three OpenAI models entered private preview in Google Cloud’s Model Garden on October 21.
- Google says Model Garden now hosts more than 200 first-party, third-party and open models.
- On OpenAI’s API, GPT-6 Astra costs $10 per million input tokens; Luna costs $0.10.
What is the OpenAI Google Cloud launch?
OpenAI put its three current frontier models and its Codex coding agent inside Google Cloud. Enterprise customers can call GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna through Google’s APIs, pay through Google Cloud Marketplace, and burn down existing Google spending commitments. Access is limited to a private preview for now.
The OpenAI Google Cloud deal was announced in parallel posts by OpenAI and Google Cloud. Google lists the models as “GPT-Astra, GPT-6.1 Sol, and GPT-6 Luna.”
Google says they are “rolling out now in private preview to select enterprise customers in Google Cloud’s model garden, with more coming soon.” OpenAI puts general availability “in the coming weeks.” Everyone else fills out an interest form.
What works today
One piece is live now. Google writes that “starting today, customers can build with these models through our APIs and connect their local Codex CLI and Desktop workflows directly to our Gemini platform for inference.”
That means a developer can keep OpenAI’s coding agent and send the compute bill to Google. For large companies with a committed Google budget, that is the real hook.
What is still coming
OpenAI says its models will “soon be available” in Google Antigravity, Google’s agent-first coding environment, and in the Gemini Agent. Google’s post phrases the second destination as “Gemini for business.” Neither post attaches a date.
How much do OpenAI models cost on Google Cloud?
Nobody has said. Neither the OpenAI post nor the Google Cloud post includes a per-token price. The only public benchmark is OpenAI’s own API list price, which ranges from $0.10 per million input tokens for Luna to $10 for Astra. Marketplace pricing could match it, or not.
Here is what the same three models cost when bought directly from OpenAI, according to the pricing table in OpenAI’s GPT-6.1 Sol launch post.
| Model | Input ($/1M tokens) | Cached input ($/1M) | Output ($/1M) | Google’s description |
|---|---|---|---|---|
| GPT-6 Astra | $10.00 | $1.00 | $50.00 | Demanding coding, computer use, science |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 | “Near-Astra intelligence at lower cost” |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 | Fast, cost-efficient everyday work |
| Source: OpenAI API list prices. Google Cloud Marketplace prices not disclosed as of October 21, 2026. | ||||
The spread is 100x on input between the top and bottom of the lineup. That matters because Google is pitching routing, not one model.
“Modern enterprise AI requires matching each workload to the right model rather than relying on a single model for every task,” wrote Michael Gerstenhaber, Google Cloud’s vice president of product management for Gemini.
OpenAI’s own numbers support the routing argument. It says GPT-6.1 Sol matches Astra on its DeepSWE v1.1 software-engineering test at roughly one-fifth the cost. We covered the original Sol and Luna pricing when GPT-6 Sol landed at $2 per million tokens.
The billing detail that matters
Google says “all usage appears on a single Google Cloud invoice, draws down existing cloud commitments, and counts toward Flexible Savings Plans (FSP) across model providers.”
For a procurement team, that removes a vendor contract. A company that already signed a multiyear Google commitment can now spend part of it on OpenAI without a new negotiation.
Why is OpenAI selling through a rival’s cloud?
Distribution. Enterprise AI budgets increasingly flow through cloud marketplaces, and OpenAI now sits on all three of the largest U.S. clouds. Google Cloud was the last gap. The trade-off is that the cloud provider typically takes a cut of every sale made on its platform, Reuters reported.
The path here was fast. According to The Next Web, AWS began hosting OpenAI models in April after Microsoft’s exclusive license ended, and OpenAI added its agents to Amazon Bedrock in September.
Greg Brockman, OpenAI’s president and co-founder, framed the deal as “making our most capable models and Codex available where customers are already building.” OpenAI’s post also states that “Microsoft remains OpenAI’s primary cloud partner.”
The Anthropic precedent
The clearest evidence that cloud channels move real money comes from OpenAI’s main rival. Anthropic made 47% of its 2025 revenue through Amazon’s and Google’s cloud platforms, according to its IPO filing as reported by Reuters and cited by PYMNTS.
That is the playbook OpenAI is copying. Anthropic’s filing has dominated AI money news this month; we tracked how Anthropic’s IPO valuation talk fell to $1.5 trillion.
How does OpenAI’s cloud footprint compare now?
OpenAI is now reachable on Microsoft Azure, Amazon Web Services and Google Cloud. Google, in turn, claims to be “the only hyperscale cloud provider offering access to all three leading closed source LLMs,” meaning Gemini, Claude and GPT under one roof.
Here is the state of play, based on the announcements and The Next Web’s reporting:
- Microsoft Azure: still OpenAI’s “primary cloud partner,” per OpenAI.
- Amazon Web Services: OpenAI models since April 2026; agents on Bedrock since September.
- Google Cloud: three models in private preview from October 21; general availability “in the coming weeks.”
- Google Antigravity and Gemini Agent: “soon,” with no date.
Google also promises governance features aimed at regulated buyers. It says customer prompts and outputs “remain entirely within your designated data perimeter and are never used to train foundation models.”
Read the fine print, though. Zero Data Retention options are “coming soon.” So is Confidential Compute. A bank evaluating this today gets a preview without two of the controls Google advertises.
Who wins and who loses from OpenAI on Google Cloud?
Google Cloud wins most clearly: it collects infrastructure and marketplace revenue whichever model a customer picks. OpenAI wins reach without building new sales relationships. Gemini loses a captive audience, and Microsoft loses another slice of its old exclusivity premium.
Winners
Alphabet. Google Cloud revenue growth accelerated to about 82% in the second quarter, according to CNBC figures cited by The Next Web. Every OpenAI token sold in Model Garden still runs on Google infrastructure and lands on a Google invoice.
Mike Leone, an analyst at Moor Insights & Strategy, told Axios: “For Google, I read the deal as confidence.” He argued that once every cloud carries the top models, the edge goes to whoever best handles a company’s data, governance and costs.
OpenAI. It gets in front of Google’s enterprise base without a separate sales motion, and it gets paid from budgets that were already committed.
Losers, or at least pressured
Gemini. Leone put it bluntly: “Gemini now has to win those customers with OpenAI sitting right next to it.” Google just shipped Gemini 4 Argon; we covered how Argon led 12 of 18 benchmarks. It now competes in its own store.
Microsoft. Azure keeps the “primary partner” label, but it is no longer the only door to OpenAI. That weakens one reason enterprises chose Azure.
OpenAI’s margins, potentially. Here is the skeptical read. If the cloud takes a percentage of marketplace sales, as Reuters reports is typical, every dollar routed through Google is worth less to OpenAI than a dollar sold direct. Volume may compensate. No one has published the split.
What to watch next
The next hard numbers arrive within days. All three cloud giants report quarterly results next week, and general availability on Google Cloud is due “in the coming weeks.” Pricing disclosure is the signal that matters most.
- October 28, 2026: Alphabet and Microsoft earnings. Both companies have confirmed the date. Listen for Google Cloud backlog and any comment on third-party model revenue, and for how Microsoft describes Azure’s OpenAI position.
- October 29, 2026 (expected): Amazon earnings. The date is an estimate, not yet confirmed. Watch for Bedrock commentary on OpenAI usage since April.
- General availability on Google Cloud. Promised “in the coming weeks.” The real signal is whether Marketplace prices match OpenAI’s API list prices.
- Antigravity and Gemini Agent integration. No date yet. The moment OpenAI models appear inside Google’s own agent products is when the Gemini-versus-GPT fight turns direct.
For context on the Gemini Agent itself, see our report on the Gemini Agent launch and its 1 trillion-token clients.
FAQ
Which OpenAI models are on Google Cloud?
GPT-6 Astra (listed by Google as GPT-Astra), GPT-6.1 Sol and GPT-6 Luna, in private preview in Model Garden. Google says “more coming soon.”
Can anyone use OpenAI models on Google Cloud today?
No. Access is limited to select enterprise customers in private preview. Others can request access through Google’s interest form. General availability is planned “in the coming weeks.”
How much do OpenAI models cost on Google Cloud?
Pricing has not been disclosed. On OpenAI’s own API, input costs $10 per million tokens for Astra, $2 for GPT-6.1 Sol and $0.10 for Luna.
Can I use my Google Cloud commitment to pay for OpenAI?
Yes. Google says usage appears on a single Google Cloud invoice, draws down existing commitments and counts toward Flexible Savings Plans.
Does this end OpenAI’s relationship with Microsoft?
No. OpenAI’s announcement states that “Microsoft remains OpenAI’s primary cloud partner.” Google Cloud joins Azure and AWS as a sales channel.
Does Google train on prompts sent to OpenAI models?
Google says prompts, context and outputs stay inside the customer’s data perimeter and are never used to train foundation models. Zero Data Retention options are “coming soon.”
The bottom line
The OpenAI Google Cloud launch is a distribution win for both sides and a pricing mystery for buyers. Google gets to sell every major closed model and keep the invoice. OpenAI gets reach into committed enterprise budgets.
But today’s launch is a private preview with no published price and two key security controls still “coming soon.” For investors, the meaningful test is next week’s cloud earnings, not today’s press release. For buyers, wait for general availability and a price list before moving workloads.
Free guide · 12 pages
AI Tools Every Investor Should Use in 2026
10 tools for research, stock signals, charts and crypto, what each one costs, three ready-made stacks from $0 a month and 7 copy-paste prompts for filings and earnings calls. Enter your email and we will send it to you.
Plus one email a week: AI Money This Week, every Sunday. Unsubscribe anytime.
Sources
- OpenAI: OpenAI models are coming to Google Cloud (October 21, 2026)
- Google Cloud: Google Cloud expands model choice and flexibility, bringing OpenAI models to customers (October 21, 2026)
- OpenAI: Introducing GPT-6.1 Sol (API pricing)
- Axios: OpenAI brings its frontier models to rival Google Cloud
- The Next Web: OpenAI brings GPT-6 Astra and Codex to Google Cloud
- PYMNTS: OpenAI Brings Its Frontier AI Models to Google Cloud
Wealth Engine researches and drafts with AI tools and checks every figure against the sources above. How we report.
Leave a Reply