HAS Platform AI Agent (Yaga) AI-First Pentest AI-Native Pentest Integrations
Services Partners Blog About Us
Login
Português English Español
YagaBench

Yaga Benchmark

In AI pentesting, the same model performs far better
with Yaga's harness than on its own.

Yaga vs. isolated models

Yaga's results

Using Yaga's harness combined with several models, the results are far superior to those of any isolated model.

Yaga's results Line chart comparing the percentage of scenarios solved by Yaga (combined models) and by GPT-5.6, Opus 4.8, Grok 4.5 and Sonnet 5 on their own, in black box, gray box and white box modes. Yaga stays well above every isolated model in every mode. 0 20 40 60 80 100 Black box Gray box White box 91.2 94.6 93.5 Yaga GPT-5.6 Opus 4.8 Grok 4.5 Sonnet 5
GPT-5.6 Opus 4.8 Grok 4.5 Sonnet 5 Yaga · 4 models combined

Share of the fixed scenario set solved by test mode. Hover the points for the exact value.

Comparison

Yaga vs. each isolated model

With Yaga, your pentest is far superior to that of any model running in isolation.

Model Black box Gray box White box
Yaga · 4 models combined 91.2% 94.6% 93.5%
GPT-5.6 40.5% 49.5% 61.9%
Opus 4.8 39.7% 48.2% 61.8%
Grok 4.5 39.5% 47.6% 59.1%
Sonnet 5 28.4% 37.5% 50.6%

HackerSec internal benchmark over a fixed set of scenarios, measuring confirmed and exploitable vulnerabilities. Each model is evaluated on its own, without Yaga's harness; the combined line is production Yaga, with the four models in a single orchestrated run.

Yaga v2.6 · Last updated · July 15, 2026

See Yaga working on your target

Put the same harness to work on your scope, with HackerSec human oversight.

Explore Yaga Go to the platform