The gap between open and closed AI models in cyber capabilities keeps shrinking
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.
Vals AI’s CyberOffense benchmark tests solo agents on simulated corporate networks. OpenAI’s restricted GPT-6 Sol Daybreak scored 80.8% of objectives at about $22 per run; open-weight DeepSeek V4.1 Flash reached 72.8% for roughly $3. On the harder Lighthouse test, Daybreak completed 27 of 31 objectives in under three hours; DeepSeek managed 21. Vals says public weights cannot be restricted, short runs understate open models that progress mainly after 100 million tokens, and real-world attacks may rise.
Continue in the app — vote & join in ➔Source: The Decoder (AI) · via ahirlevel.hu