FrontierCyber
We show this table for reference; we do not rank on it.
Independent evaluation of AI agents against vulnerable real-world systems in dynamic environments.
Benchmark score on FrontierCyber — September 27, 2026
We compile the FrontierCyber rows from benchmark-owner or independent runs. GPT-6 Astra leads the table at 38.1%, followed by GPT-5.6 Sol (9.6%). We do not use these results to rank models overall.
GPT-6 Astra
OpenAI
GPT-5.6 Sol
OpenAI
2 modelsAgenticCurrentDisplay onlyUpdated September 27, 2026
Benchmark score table (2 models)
ScoreAbout FrontierCyber
Year
2026
Tasks
197 dynamic cyber tasks
Format
Tasks solved
Difficulty
Easy through elite cyber operations
FrontierCyber tests autonomous cyber performance on live, realistic systems and reports solved tasks across difficulty bands. BenchLM stores the independent snapshot as display-only external evidence.
Freshness and provenance
Version
FrontierCyber 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does FrontierCyber measure?
Independent evaluation of AI agents against vulnerable real-world systems in dynamic environments.
Which model scores highest on FrontierCyber?
GPT-6 Astra by OpenAI currently leads with a score of 38.1% on FrontierCyber.
How many models are evaluated on FrontierCyber?
2 AI models have been evaluated on FrontierCyber on BenchLM.
Compare top models on FrontierCyber
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 2,000+ readers.
One email each week. Unsubscribe anytime.