# GPT-5.6 Cyber ![Organization](https://img.shields.io/badge/Org-OpenAI-412991) ![Access](https://img.shields.io/badge/Access-Daybreak%20Red-b31b1b) ![Built on](https://img.shields.io/badge/Built%20on-GPT--5.6%20Sol-1f6feb) ![License](https://img.shields.io/badge/License-Proprietary-d29922) ![Released](https://img.shields.io/badge/Released-Aug%202026-555) > **Quick answer:** GPT-5.6 Cyber is OpenAI's purpose-trained cybersecurity model, built on GPT-5.6 Sol and released August 10, 2026 as part of the expanded Daybreak program's Red access tier. It's trained to reduce refusals on dual-use security tasks — completing 95.0% of requests on OpenAI's internal Advanced Cybersecurity Completion Rate evaluation, versus 57.3% for the predecessor GPT-5.5 Cyber and just 1.5–2.0% for GPT-5.6 Sol under standard safeguards. ## At a Glance **Where it leads** - Highest completion rate on advanced, dual-use cybersecurity tasks among OpenAI's cyber-tier models: 95.0% vs 57.3% for GPT-5.5 Cyber - Used to discover real zero-day vulnerabilities in production software, including CVE-2026-15903 (a high-severity V8/Chrome JIT compiler bug) - Outperforms GPT-5.5 Cyber on every internal cyber evaluation OpenAI published (ExploitGym, Zero-Day Discovery, Vulnerability Discovery and Report Writing, ExploitBench) **Where it lags** - Scores worse than GPT-5.6 Sol (Daybreak Blue) on the Vulnerability Discovery and Report Writing eval — attributed to producing shorter, less detailed reports - Trails GPT-5.6 Sol (Daybreak Blue) on ExploitBench under the standard 300-turn limit, though the gap narrows at 600 turns - Not broadly available — access is gated behind identity verification, account security requirements, and legal attestations via Daybreak Red - Rated High (not Critical) for cybersecurity capability under OpenAI's Preparedness Framework — the same threshold as the base GPT-5.6 Sol model **Best for:** Vetted security researchers and defenders conducting authorized vulnerability research, exploit validation, and advanced penetration testing that base models refuse even with reduced safeguards. ## What GPT-5.6 Cyber Is GPT-5.6 Cyber is a specialized fine-tune of [GPT-5.6 Sol](/models/openai/gpt-5-6-sol), OpenAI's flagship model, trained specifically to improve performance on cybersecurity workflows involving exploit development and advanced security research, and to reduce refusals on certain dual-use tasks (e.g. exploit-chain development, authentication bypass, privilege escalation) that GPT-5.6 Sol will decline even when run with reduced "Daybreak Blue" safeguards. It's distributed through **Daybreak**, OpenAI's program for giving vetted defenders frontier cybersecurity capability before attackers can weaponize similar capabilities at scale. Daybreak has two access tiers: **Daybreak Blue** provides frontier general-purpose models (GPT-5.6 Sol) with safeguards tailored to authorized defensive work; **Daybreak Red** — which GPT-5.6 Cyber requires — provides purpose-trained cybersecurity models for authorized vulnerability research, exploit validation, and security testing. Both tiers require identity verification, account security controls, monitoring, and legal attestations; individual Daybreak accounts will be required to adopt hardware security keys starting September 1, 2026. Despite its specialized training, GPT-5.6 Cyber does not cross OpenAI's Preparedness Framework "Critical" threshold for cybersecurity capability — it's assessed at "High," the same tier as GPT-5.6 Sol. OpenAI has stated GPT-5.6 Cyber was not involved in the separately-disclosed Hugging Face security incident. ## Specifications | Field | Value | |---|---| | Organization | OpenAI | | Built on | GPT-5.6 Sol | | Access tier | Daybreak Red (trusted/vetted defenders only) | | License | Proprietary, limited access | | Release date | August 10, 2026 | | Preparedness capability rating (cybersecurity) | High (below Critical threshold) | ## Access & Pricing GPT-5.6 Cyber has no public API pricing — it is not available through standard OpenAI API access. Approved individuals and organizations conducting authorized security work can apply for Daybreak Red access at [openai.com/daybreak/partners](https://openai.com/daybreak/partners/). Access requires identity verification, account security controls (hardware security keys required from September 1, 2026), usage monitoring, approved-use restrictions, and legal attestations. ## Notable In-House Benchmark Results (Not on Public Leaderboards) > The evaluations below are OpenAI-authored internal benchmarks described in the [Daybreak expansion announcement](https://openai.com/index/expanding-daybreak-as-the-cyber-defense-window-narrows/), without public datasets or independent third-party verification — comparable to a vendor's internal red-team eval rather than a community benchmark like SWE-Bench. They are not added to Benchgen's public leaderboards for this reason, per Benchgen's convention for single-vendor, non-reproducible evaluations. The Advanced Cybersecurity Completion Rate values are stated exactly in OpenAI's post; the remaining four evaluations' values are read directly off OpenAI's published bar charts and should be treated as approximate (±a few percentage points). **Advanced Cybersecurity Completion Rate** — how often each model responds (rather than refuses) to advanced, dual-use cybersecurity prompts involving exploit-chain development, authentication bypass, and privilege escalation: | Model / access tier | Completion rate | |---|---| | GPT-5.6 Sol (safeguards enabled) | 1.5% | | GPT-5.6 Sol (Daybreak Blue) | 2.0% | | GPT-5.5 Cyber (Daybreak Red) | 57.3% | | **GPT-5.6 Cyber (Daybreak Red)** | **95.0%** | **ExploitGym — Intended vulnerability** (success rate turning a known vulnerability into a working exploit achieving arbitrary code execution, controlled environment; chart-estimated): | Model / access tier | Success rate | |---|---| | GPT-5.6 Sol (safeguards enabled) | ~0% | | GPT-5.5 Cyber (Daybreak Red) | ~21% | | GPT-5.6 Sol (Daybreak Blue) | ~31% | | GPT-5.6 Cyber (Daybreak Red) | ~32% | **Zero-Day Discovery Eval** (average discovery quality score on a proprietary internal scale, rewarding severity/impact of found vulnerabilities plus write-up quality and calibration; chart-estimated): | Model / access tier | Avg. discovery score | |---|---| | GPT-5.6 Sol (safeguards enabled) | ~0.00 | | GPT-5.6 Sol (Daybreak Blue) | ~0.16 | | GPT-5.5 Cyber (Daybreak Red) | ~0.18 | | GPT-5.6 Cyber (Daybreak Red) | ~0.24 | **Vulnerability Discovery and Report Writing** (score on a proprietary internal scale for finding, exploiting, and writing up vulnerabilities in a repo; chart-estimated): | Model / access tier | Score | |---|---| | GPT-5.6 Sol (safeguards enabled) | ~0.00 | | GPT-5.5 Cyber (Daybreak Red) | ~0.12 | | GPT-5.6 Cyber (Daybreak Red) | ~0.15 | | GPT-5.6 Sol (Daybreak Blue) | ~0.18 | **ExploitBench** (capability coverage developing a V8 vulnerability into a full exploit, harder than ExploitGym due to active sandbox defenses; chart-estimated): | Model / access tier | 300-turn limit | 600-turn limit | |---|---|---| | GPT-5.5 Cyber (Daybreak Red) | ~73% | ~81% | | GPT-5.6 Cyber (Daybreak Red) | ~73% | ~84% | | GPT-5.6 Sol (Daybreak Blue) | ~76% | ~84% | GPT-5.6 Sol (Daybreak Blue) is more token-efficient and edges out GPT-5.6 Cyber on ExploitBench specifically under the standard 300-turn budget; the gap narrows once agents are allowed 600 turns. ## Real-World Vulnerability Discovery Beyond internal benchmarks, OpenAI used GPT-5.6 Cyber to investigate real, unmodified software. It uncovered two previously unknown vulnerabilities in V8 (Chrome's JavaScript engine) that could be chained to corrupt memory and escape the V8 heap sandbox; these were responsibly disclosed to Google and fixed as **CVE-2026-15903**, a high-severity bug where the optimizing JIT compiler skipped a safety check during integer conversion, enabling out-of-bounds memory access. OpenAI also reports GPT-5.6 Cyber found: at least five vulnerabilities in a popular mobile OS (including an untrusted-app-to-local-privilege-escalation chain), three critical vulnerabilities in a popular database (including a remote code execution path), and over 400 privilege-escalation vulnerabilities in a popular OS kernel — all under coordinated disclosure with the relevant vendors as of this writing. ## GPT-5.6 Cyber vs Alternatives | Model | Advanced Cyber Completion Rate | Access tier | License | |---|---|---|---| | **GPT-5.6 Cyber** | **95.0%** | Daybreak Red | Proprietary, limited | | [GPT-5.6 Sol](/models/openai/gpt-5-6-sol) (Daybreak Blue) | 2.0% | Daybreak Blue | Proprietary | | [GPT-5.5 Cyber](/models/openai/gpt-5-5-cyber) | 57.3% | Daybreak Red (predecessor) | Proprietary, limited | GPT-5.6 Cyber's value isn't raw capability over GPT-5.6 Sol — the two share the same underlying model — it's refusal reduction on legitimate, dual-use security work that a general-purpose safeguarded model won't perform, paired with specialized training that improves severity calibration and exploit-chain completion on tasks OpenAI directly trained for. ## Access and Safety Because Daybreak models run with reduced or removed safeguards, OpenAI recommends several operational practices for teams using GPT-5.6 Cyber: - **Sandbox and isolate.** Run security workflows in controlled environments without access to sensitive production systems or the open internet. - **Use auto-review mode in Codex.** Auto-review evaluates tool calls requiring elevated permissions before execution and can block actions with a significant risk of destructive behavior — OpenAI is actively encouraging Daybreak/Codex users to switch from full-access mode to auto-review. - **Define and enforce scope.** Use scoped permission profiles to specify which systems and actions are authorized for the agent. - **Hardware security keys required from September 1, 2026** for all individual Daybreak accounts. ## Frequently Asked Questions

What is GPT-5.6 Cyber?

GPT-5.6 Cyber is OpenAI's purpose-trained cybersecurity model, built on GPT-5.6 Sol and released August 10, 2026 through the Daybreak Red access tier. It's trained to improve exploit-development and vulnerability-research capability while reducing refusals on legitimate, dual-use defensive security tasks.

How is GPT-5.6 Cyber different from GPT-5.6 Sol?

They share the same underlying model, but GPT-5.6 Cyber is fine-tuned to complete far more advanced/dual-use cybersecurity requests (95.0% vs 1.5–2.0% for GPT-5.6 Sol on OpenAI's internal Advanced Cybersecurity Completion Rate eval) and to improve on specific exploit-development and zero-day-discovery tasks. GPT-5.6 Sol still edges out GPT-5.6 Cyber on the internal Vulnerability Discovery and Report Writing eval and on ExploitBench under a 300-turn limit.

How does GPT-5.6 Cyber compare to GPT-5.5 Cyber?

GPT-5.6 Cyber outperforms GPT-5.5 Cyber (its predecessor) on every internal cyber evaluation OpenAI published, most notably completing 95.0% of advanced cybersecurity requests versus 57.3% for GPT-5.5 Cyber — directly addressing feedback from security researchers about persistent refusals in the earlier model.

Can anyone access GPT-5.6 Cyber?

No. GPT-5.6 Cyber is only available through Daybreak Red access, granted to vetted individuals and organizations conducting authorized vulnerability research, exploit validation, or security testing. Access requires identity verification, account security controls including hardware security keys (required from September 1, 2026), monitoring, and legal attestations.

Has GPT-5.6 Cyber found real vulnerabilities?

Yes. OpenAI used it to discover CVE-2026-15903, a high-severity V8/Chrome JIT compiler vulnerability, along with vulnerabilities in a popular mobile OS, a popular database, and over 400 issues in a popular OS kernel — all under coordinated disclosure with the affected vendors.

--- *Scores and details sourced from OpenAI's ["Expanding Daybreak as the Cyber Defense Window Narrows"](https://openai.com/index/expanding-daybreak-as-the-cyber-defense-window-narrows/) announcement, August 10, 2026. Chart-derived figures are approximate. Last updated 2026-08-11.*

GPT-5.6 Cyber — Training jobs

Training jobs