OpenAI launches GPT-6 Cyber at higher API price than Anthropic, Google
For the CISO of a regional US bank
OpenAI released GPT-6 Cyber, a specialized security model, initially to selected customers through its gated Daybreak Red program, VentureBeat reported; the report dates the release only as "today".1
OpenAI describes GPT-6 Cyber as a variant of GPT-6 Astra with additional training for security investigations, and its system card says reinforcement learning rewards reproducing crashes and constructing exploits, alongside behavior that respects authorization.1 The report does not say why OpenAI set the price above its predecessor and rivals.1
OpenAI rates the model's cybersecurity capability Critical under its Preparedness Framework, up from High for GPT-5.6 Cyber.1
OpenAI said researchers using GPT-6 Cyber reproduced 1,724 Linux kernel vulnerabilities and developed working exploits for 948, while GPT-5.6 Sol produced exploits for 166 in the same setup; VentureBeat notes the 948 should not be read as a count of previously unknown or unpatched flaws.1
OpenAI's developer documentation lists $20 per million input tokens and $100 per million output tokens, which VentureBeat calculates as twice Anthropic's Claude Mythos 5.1 ($10 and $50) and five times Google's Gemini 4 Argon Cyber Permissive ($4 and $20).1
Daybreak Red applicants must upload information OpenAI uses for review, such as the defensive workflows they plan to run; the performance figures are OpenAI's own, and VentureBeat says the announcement does not provide an independent comparison across vendors.1
Banks and fintechs that want AI-driven vulnerability research will pay more per token for GPT-6 Cyber and must pass a workflow review first, so they should compare cost per validated finding rather than list price. Vendor-reported benchmarks are not independent evidence, so pilots on their own code are the safer test.
OpenAI plans to extend GPT-6 Cyber access across the Daybreak Red tier and did not give a date.1
Our take. OpenAI's GPT-6 Cyber wrote working exploits for 948 Linux kernel flaws and sits behind an application form at $100 per million output tokens. The form slows your bank. Whether it slows attackers for long is the bet.
Action. Security operations: Submit a Daybreak Red application naming defensive workflows, and budget token spend at $20 input and $100 output per million tokens.
What our records say
OpenAI sells banks ChatGPT seats and API access and in 2026 added on-site engineering deals, a financial-services edition and a separately funded deployment company, while its bank customers include Morgan Stanley, BNY and Customers Bank, the deepest tie.1 Among the 100 banks tracked, Morgan Stanley's advisor assistants run on OpenAI models, BNY builds agents on them, and Customers Bank, SouthState, JPMorganChase and Goldman Sachs use OpenAI tools or models.1
From our records
- ChatGPT Enterprise (OpenAI)vendor · banks using it (14) Morgan Stanley, BNY, Customers Bank · alternatives Microsoft Copilot, Claude, Gemini Enterprise
- Claude (Anthropic)vendor · banks using it (21) Goldman, JPMorgan, BankUnited · alternatives Microsoft Copilot, ChatGPT Enterprise, Gemini Enterprise
- Gemini Enterprise (Google Cloud)vendor · banks using it (9) Wells Fargo, BNY, Citi · alternatives Microsoft Copilot, ChatGPT Enterprise, Claude
Sources
- ReportOpenAI launches GPT-6 Cyber at higher API price than Anthropic, Google Venturebeat