Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

LLM Red-teaming research benchmarks · SOTA2 Research

LLM Red-teaming

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

S-GFN-defended Target Model

7.33Unsuccessful Attack Rate (UA)

9

May 4, 2026

Rainbow Teaming defended Target Model

0UA

9

May 4, 2026

Jailbreak R1-defended Target Model

87.67UA

9

May 4, 2026

GFN-defended Target Model

0.33Unsuccessful Attack Rate (UA)

9

May 4, 2026

Target Victim Model

2Unknown/Unsafe Attacks

9

May 4, 2026

WildTeaming Target: Mistral

55.5ASR

2

Apr 24, 2026

WildTeaming Llama-3

25.4ASR

2

Apr 24, 2026

WildTeaming Llama-3.3-70B target

Adaptive Instruction Composition

3,711Effectiveness Count

2

Apr 24, 2026

WildTeaming Mistral-7B target

Adaptive Instruction Composition

5,414Effectiveness Query Count

2

Apr 24, 2026

Popular datasets

WildTeaming Jailbreak Target Model Target Victim Model Rainbow Teaming

Follow for updates

Join SOTA2 Contact