FrontierCyber
Agents look for and exploit zero-day vulnerabilities in current off-the-shelf software and hardware at four difficulty levels. Scored as challenges solved.
Irregular published the v1.0 design on 22 June 2026. Two subsets run today, mobile devices and software exploitation, with named targets including Pillow, lxml, FFmpeg, ImageMagick, PostgreSQL, MongoDB, Redis, phones and routers. Each challenge gets a predicted difficulty band, Easy to Elite, before any run. A hidden flag proves success and planted markers give partial credit. Each evaluation ties to a timestamped snapshot, so scores compare only within one snapshot. Not public.
- Level
- Research frontier Researchers, on open problems or tasks that take days
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
GPT-6 Astra leads with 38.1%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 GPT-6 Astra 38.1%
- 2 GPT-5.6 Sol 9.64%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra | 38.1% | Sep 2026 | Model card | |
| 2 | GPT-5.6 Sol | 9.64% | Jun 2026 | Model card |
2 models have held number one since its release in Jun 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
3 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 3 Sep 2026 | GPT-6 Astra system card | 38.1 | Irregular FrontierCyber, 86 of 226 challenges solved (10.1.2.6) | Cyber risk |
| 9 Jul 2026 | GPT-5.6 Sol system card | 9.64 | Irregular FrontierCyber, 19 of 197 challenges solved (Easy 11%, Medium 12%, Hard 5%, Elite 0%) (Figure 36, 9.1.2.5) | Cyber risk |
| 25 Jun 2026 | GPT-5.6 Sol system card | 9.64 | Irregular FrontierCyber, 19 of 197 challenges solved (Easy 11%, Medium 12%, Hard 5%, Elite 0%) (Figure 36, 9.1.2.5) | Cyber risk |
More in Cybersecurity
- ExploitGym In use 869 real-vulnerability exploit-development tasks (userspace, V8, Linux kernel). Success means capturing a flag through the intended vulnerability. GPT-6 Astra 42.4%
- ExploitBench In use 41 recent V8 JavaScript-engine N-day vulnerabilities. Progress measured by 16 grader-verified capability flags across five tiers. GPT-6 Astra 100%
- CWE-bench In use Agents audit a repository and write a patch for a security flaw, without being told where the bug is. Covers six programming languages.
Gemini 4 Carbon 71% - Real-world vulnerability discovery In use Agents search popular open-source codebases to rediscover about 1,200 recent confirmed vulnerabilities across many languages. Reports share discovered. Gemini 4 Argon 86%
- Artificial Analysis Cyber Index In use Agents do enterprise cyber defense tasks from three benchmarks, CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA, combined into one success rate.
Mistral Large 4 (Le Chonk, open weights) 51% - Big Sleep Evaluation (Google) In use Agents search large real codebases such as Chrome and Safari for critical, hard to find vulnerabilities. Scored by pass@1 success rate, run without safety guardrails. Gemini 3.5 Flash Cyber 72%