Vals AI published an updated version of its CyberOffense benchmark on October 23, 2026, measuring models’ ability to conduct multi-step offensive operations. On the simulated Lighthouse network, OpenAI’s GPT-6 Sol completed 27 of 31 objectives, while the open DeepSeek V4.1 Flash 1017 model achieved 21. DeepSeek’s model cost less than $3 per attempt, compared with about $22 for GPT-6 Sol, which is only available through OpenAI’s Daybreak program. The findings indicate that restricted access to the most advanced models provides an increasingly limited advantage in performance and cost, particularly for attackers who can repeatedly run cheaper models.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
