OpenAI Prices GPT-6 Cyber at 2x Astra, Behind Daybreak Red

OpenAI’s API model page lists GPT-6-Cyber at $20 per million input tokens and $100 per million output tokens, and says “Your organization needs Daybreak Red access and approval to use GPT-6-Cyber.” (OpenAI)

OpenAI’s models page lists GPT-6 Astra, which it describes as “Our most capable model for the most demanding work”, at $10 per million input tokens and $50 per million output tokens (OpenAI). By our arithmetic, the cyber model costs twice as much on both input and output.

Business Pill · AI Agents Explained

A short explainer on AI agents, systems that move from answering to doing. It teaches the general idea only and does not discuss this story.

The key insight: OpenAI lists GPT-6-Cyber at $20 and $100 per million input and output tokens, twice GPT-6 Astra’s $10 and $50 by our arithmetic, and requires Daybreak Red access and approval to use it.

The Price and the Gate

The GPT-6-Cyber page lists cached input at $2.00 per million tokens and cache writes at $25.00, and notes: “These are standard short-context prices.” It lists a 1,050,000-token context window and 128,000 maximum output tokens.

Access is not self-serve. The page says “Your account team can confirm access” and that “The standard rate limits below apply to approved organizations and may differ if your organization has custom limits.”

OpenAI's listed standard short-context prices per million tokens for GPT-6 Astra and GPT-6-Cyber. Source: Open
OpenAI’s listed standard short-context prices per million tokens for GPT-6 Astra and GPT-6-Cyber. Source: OpenAI API documentation, read 7 November 2026. Ratio: ours.

What OpenAI Says It Used the Model For

In its launch post, OpenAI writes: “In controlled red-teaming campaigns, agents using GPT‑6 Cyber and other OpenAI models chained vulnerabilities to escape microVMs and execute code on the underlying hosts.” It adds that it used the findings to “patch software, tighten network restrictions, and report issues to maintainers and vendors” (OpenAI).

OpenAI credits that result to agents using GPT-6 Cyber together with other OpenAI models, not to GPT-6 Cyber alone. Our earlier coverage reported the model’s benchmark results, including 48.5% on ExploitGym, up from 32.1% (our piece).

Our Numbers

By our arithmetic: $20 against $10 per million input tokens and $100 against $50 per million output tokens is a 2x ratio on both. A job using one million input tokens and one million output tokens would cost $120 on GPT-6-Cyber and $60 on GPT-6 Astra at the listed short-context rates, before any cached-input discount.

Connecting the Coverage

The pricing sits beside other cyber-specific moves this week: an industry scale for rating cyber jailbreaks backed by OpenAI and four rivals (our piece) and a Varonis proof of concept that runs a command loop through Amazon Bedrock inside a breached AWS account (our piece). OpenAI’s earlier cyber model, gpt-5.4-cyber, was given 20 days’ notice of shutdown (our piece).

The Structural Read

As we read the model page, the cyber model is priced above the model OpenAI calls its most capable, and it is sold only to approved organisations.

As we read the launch post, OpenAI frames the model’s red-teaming results as work by agents using GPT-6 Cyber and other OpenAI models, not the cyber model alone.

As we read the pricing, a buyer pays for access in two ways: a higher per-token rate and an approval step that the page says an account team confirms.

OpenAI, GPT-6-Cyber model page

“Your organization needs Daybreak Red access and approval to use GPT-6-Cyber.”

Three Implications

FOR SECURITY TEAMS BUDGETING AI At OpenAI’s listed rates, one million input tokens and one million output tokens cost $120 on GPT-6-Cyber against $60 on GPT-6 Astra, by our arithmetic.

FOR ACCESS OpenAI’s page says approved organisations get standard rate limits that may differ under custom limits; it does not publish the approval criteria.

FOR READING THE RED-TEAMING CLAIM OpenAI attributes the microVM escapes to agents using GPT-6 Cyber and other OpenAI models in controlled campaigns.

The Business Engineer Lens

This story maps onto the Business Engineer framework The CFO’s Guide to the Token Economy.

The analysis says: “It is simultaneously the unit of cognition the model produces, the unit of compute the data center serves, the unit of price the lab charges, and the unit of value the enterprise extracts.”

As we read it, OpenAI’s two price points for GPT-6 Astra and GPT-6-Cyber are the same unit of price set at different levels for different buyers; the analysis does not discuss GPT-6-Cyber.

What Is Not Established

OpenAI’s pages do not say how many organisations have Daybreak Red approval, what the approval criteria are, or whether prices differ for custom agreements beyond the rate-limit note. We have not tested the model.

This Week In The AI Supercycle

Anthropic’s Reality Check, The Capital Squeeze, & The New World Order

This week’s issue: for two years the AI build-out was priced as a growth story. This week it started being priced as a credit story.

Read the full issue →

Business Engineer Framework

The CFO’s Guide to the Token Economy

The Business Engineer analysis of the token as the unit of AI economics.

Read the Map of AI →

The Bottom Line

OpenAI lists GPT-6-Cyber at $20 and $100 per million input and output tokens, twice GPT-6 Astra’s $10 and $50 by our arithmetic, and requires Daybreak Red access and approval to use it.

95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.

A note on sourcing. Prices, access terms and the red-teaming quote come from OpenAI’s API documentation and launch post, linked in the text and read on 7 November 2026; the ratio and the sample-job cost are our arithmetic.

Sources: OpenAI API: GPT-6-Cyber model page · OpenAI API: models page (GPT-6 Astra) · OpenAI: Introducing GPT-6 Cyber

Scroll to Top
FourWeekMBA