AI Model Overview

OpenAI GPT-5.6 Cyber — what it is, when to use it and what to use instead

Active

No. GPT-5.6 Cyber is active at OpenAI and has no announced end date.

gpt-5.6-cyber last verified

Summary

GPT-5.6 Cyber is OpenAI's cybersecurity model for authorised vulnerability research and security testing, at $12.5 per million input tokens and $75 per million output tokens. Access requires separate approval through OpenAI's Daybreak programme, and it is only available on the Responses API.

Cyber is a specialised branch of the GPT-5.6 line rather than a general tier. It shares the February 2026 knowledge cutoff of Sol, Terra and Luna, but uses a 400K-token context window (272K max input) instead of 1.05M, with 128K max output.

It accepts text and image input and returns text, with reasoning tokens, structured outputs, function calling, web search, code interpreter, hosted shell, computer use and MCP. Only the Responses endpoint is supported.

Access is gated: OpenAI requires an approved application to the Daybreak programme before the model can be provisioned. Plan for that lead time before you design around it. No deprecation has been announced for this model; OpenAI names it as the replacement for GPT-5.4 Cyber, which shuts down on 1 October 2026.

Modality
  • Text
  • Vision
Use case
  • Chat
  • Coding
  • Reasoning
  • Agents & tool use
  • Long context
  • Web search
Tier
  • Frontier
Access
  • Hosted API

Specifications

Input price
$12.50 / M tokens
Output price
$75.00 / M tokens
Price band
Flagship
Context window
400K tokens
Max output
128K tokens
Input
Text, image
Output
Text
Knowledge cutoff
February 2026
Tool use
Yes
Structured output
Yes
Weights licence
not published
Model maker
OpenAI

“unknown” means the figure has not been published, or we have not verified it yet — never that the answer is no. “not published” under the weights licence means this model is not offered with downloadable weights. The price band is ours: it bands the blended price (three parts input to one part output) as budget (≤ $0.50), standard (≤ $3), premium (≤ $12) or flagship.

Lifecycle

OpenAI has not announced a deprecation or shutdown date for this model. That is a statement about what has been published, not a guarantee — providers typically announce a shutdown months ahead, and this page updates when they do.

  1. Released
  2. Deprecated not announced
  3. Retired not announced

Use it for

  • Run approved vulnerability research and exploit validation as an authorised defender.
  • Support authorised security testing engagements where a general model refuses the task.

Avoid it when

  • You have not been approved through the Daybreak programme; the model cannot be provisioned without it.
  • The task is general-purpose; GPT-5.6 Sol costs $4 / $20 against $12.5 / $75 here and has a 1.05M-token window.
  • Your integration uses Chat Completions or Batch; only the Responses endpoint is supported.
  • You need a context window above 400K tokens.

Alternatives to consider

  • GPT-5.6 Sol Curated

    OpenAI Active

    The general flagship of the same generation at $4 / $20, with a 1.05M-token window and no access approval.

    $4.00 in · $20.00 out / M tokens 1.1M context

  • GPT-6 Astra Curated

    OpenAI Active

    OpenAI's most capable general model at $10 / $50, if the work does not require a security-specific model.

    $10.00 in · $50.00 out / M tokens 1.1M context

  • Claude Opus 5 Curated

    Anthropic Active

    Anthropic's top general tier at $5 / $25, available without a programme application.

    $5.00 in · $25.00 out / M tokens 1M context

  • GPT-5.2 Pro Derived

    OpenAI Active

    Same provider, same tier (frontier), similar price; shares chat, coding, reasoning, agents, long-context and search.

    $21.00 in · $168.00 out / M tokens 400K context

  • GPT-5 pro Derived

    Microsoft Azure AI Foundry Active

    Alternative from Microsoft Azure AI Foundry, same tier (frontier), similar price; shares chat, coding, reasoning, agents, long-context and search.

    $15.00 in · $120.00 out / M tokens 400K context

Curated suggestions were picked by a reviewer, with the reason written by hand. Derived suggestions are ranked from the data — shared tags, the same tier and a comparable price band — not from benchmark scores. Check the capability fit yourself: we track lifecycle, price and specifications.

Sources

Verified against the source on . Dataset snapshot . Providers can change a date without notice — the link above is always the final word.

About this profile

The specifications and dates above are facts from OpenAI, with the sources listed. The summary, the use-for and avoid-when lists and the curated alternatives are our assessment: editorial profile reviewed on by llm-researcher.

Common questions

What is GPT-5.6 Cyber for?
GPT-5.6 Cyber is OpenAI's cybersecurity model for authorised vulnerability research and security testing, at $12.5 per million input tokens and $75 per million output tokens. Access requires separate approval through OpenAI's Daybreak programme, and it is only available on the Responses API.
What does GPT-5.6 Cyber cost?
$12.50 per million input tokens and $75.00 per million output tokens, as last verified on 2026-09-10.
What is the context window of GPT-5.6 Cyber?
400K tokens.
Is GPT-5.6 Cyber deprecated?
No. GPT-5.6 Cyber is active at OpenAI and has no announced end date.
What should I use instead of GPT-5.6 Cyber?
OpenAI GPT-5.6 Sol — The general flagship of the same generation at $4 / $20, with a 1.05M-token window and no access approval. OpenAI GPT-6 Astra — OpenAI's most capable general model at $10 / $50, if the work does not require a security-specific model. Anthropic Claude Opus 5 — Anthropic's top general tier at $5 / $25, available without a programme application.

All OpenAI models and their end dates →