GPT-5.6 Cyber
Released
GPT-5.6 Cyber is an AI model released by OpenAI on Monday, Aug 10 2026, 45 days after GPT-5.6 Luna. It has a 400k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover Advanced Cybersecurity Completion Rate (Internal).
Get OpenAI releases and AI news.
Every other Monday
API pricing
- InputWhat you pay for everything you send the model — your question, plus any documents or earlier conversation you include with it.
- $12.50
- Cached inputA reduced rate for text you send over and over. If every request starts with the same instructions or the same document, the provider keeps a copy ready and charges less to read it again.
- $1.25
- Cache writeA one-off charge for storing text so later requests can reuse it at the cheaper cached rate.
- $15.625
- OutputWhat you pay for the text the model writes back. It is normally the dearer half: producing an answer costs more than reading one.
- $75.00
- Reasoning or thinking is supported.
- The API accepts at most 272K input tokens. Cache writes cost $15.625 per million tokens for 30 minutes or $31.25 for 12 hours.
Benchmarks
Compare GPT-5.6 Cyber with
Suggested comparisons
GPT-5.6 CybervsGPT-6.1 TerraGPT-5.6 CybervsClaude Fable 5.2GPT-5.6 CybervsGemini 4 CarbonGPT-5.6 CybervsMuse EmberGPT-5.6 CybervsGrok 4.7GPT-5.6 CybervsDeepSeek-V4.1-Flash-1017GPT-5.6 CybervsMistral Large 4GPT-5.6 CybervsKimi K3.1 ProGPT-5.6 CybervsGLM-5.3-FlashGPT-5.6 CybervsQwen3.8-Max-0902GPT-5.6 CybervsNemotron 3.5 Bolt E4B 4GBAbout
GPT-5.6 Cyber was introduced on August 10, 2026 as a GPT-5.6 Sol derivative for advanced, authorized cybersecurity work. OpenAI described training for vulnerability research and exploit validation while reducing refusals on certain dual-use requests. Access was offered through Daybreak Red, with identity verification, monitoring and approved-use restrictions.
OpenAI classified the model as High rather than Critical for cybersecurity under its Preparedness Framework. Its announcement reported gains on some specialized tasks, but also described evaluations where general GPT-5.6 Sol performed better. The release should therefore not be described as uniformly superior on every security task. Daybreak Blue separately provided general-purpose models with safeguards adjusted for approved defensive workflows. GPT-6 Cyber followed on November 5, 2026, and OpenAI rated that later model Critical for cybersecurity, up from High for GPT-5.6 Cyber. In that announcement OpenAI reported 32.1% for GPT-5.6 Cyber on ExploitGym, against 48.5% for GPT-6 Cyber, and said GPT-5.6 Cyber was being made available to all Daybreak Red customers while GPT-6 Cyber started with a smaller approved group.
Frequently asked questions
GPT-5.6 Cyber was released by OpenAI on Monday, Aug 10 2026.