# Vals tests put open-weight DeepSeek close to gated cyber model

_Published Friday, November 6, 2026 at 2:43 PM EST · AI, Security · Latest · Tier 2 — Notable_

![Vals tests put open-weight DeepSeek close to gated cyber model — Primary](https://media.wired.com/photos/6aebaf11899aefeeb0331a11/191:100/w_1280,c_limit/Security_Open-Weight%20AI%20Models%20Are%20About%20to%20Hit%20a%20Key%20Cyber%20Milestone_v1.jpg)

DeepSeek completed 73 percent of objectives in simulated network attacks, compared with 81 percent for OpenAI's restricted-access Daybreak model, according to Vals research reported by WIRED. The tests covered two scenarios, including compromising one company network to reach another.

Vals also reported token costs of $2 to $3 for DeepSeek attacks, compared with $17 to $46 for GPT-6-Sol. That cost comparison involves a different OpenAI model from the capability comparison. The results concern simulated networks; observed attackers often use older, less capable models.

## Sources

- [WIRED](https://www.wired.com/story/free-chinese-ai-models-are-getting-scary-good-at-hacking/)

---
Canonical: https://techandbusiness.org/newswire/EYgM8kv9ko7h9WVOqfaxDC
Published: 2026-11-06T19:43:40.876Z
Story chronology: 2026-11-06T18:43:43.052Z
Retrieved: 2026-11-06T22:56:45.393Z
Publisher: Tech & Business (techandbusiness.org)