Research · 9 November 2026
Research

Open‑weight DeepSeek V4.1 Flash achieves 72.8% cyber‑attack success at one‑tenth cost

In the Vals AI benchmark, OpenAI's GPT‑6 Sol Daybreak hit 80.8% of attack objectives, while open‑weight DeepSeek V4.1 Flash reached 72.8% at roughly one‑tenth the cost. Longer runtimes were critical for open models to demonstrate capability.

Why it matters

DeepSeek V4.1 Flash achieves 72.8% attack success at roughly one‑tenth cost; evaluate cost‑effective red‑team options versus GPT‑6.

For product managers

For your product
What does this mean for your product?
Try it on your product

Get this every morning, written for your product.

Deltaday reads the whole AI news cycle and tells you what changed, whether it's real yet, and what to do about it.