Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Deception Detection on Sandbagging (Sandbag)
Loading...
99.9
AUROC
STATEWITNESS
56.844
68.022
79.2
90.378
Jun 16, 2026
AUROC
Updated 1mo ago
Evaluation Results
Method
Method
Links
AUROC
STATEWITNESS
Target Model=GPT-OSS-20B
2026.06
99.9
LLM-as-a-Judge
Target Model=GEMMA4-26...
2026.06
99.3
STATEWITNESS
Target Model=GEMMA4-26...
2026.06
99.2
Self-Evaluation
Target Model=GEMMA4-26...
2026.06
99.1
LLM-as-a-Judge w/ CoT
Target Model=GPT-OSS-20B
2026.06
98.9
LR probe
Target Model=GPT-OSS-20B
2026.06
98.4
LLM-as-a-Judge w/ CoT
Target Model=GEMMA4-26...
2026.06
98.4
CMMS probe
Target Model=GPT-OSS-20B
2026.06
96.8
CMMS probe
Target Model=GEMMA4-26...
2026.06
94.6
Self-Evaluation
Target Model=GPT-OSS-20B
2026.06
93.9
LR probe
Target Model=GEMMA4-26...
2026.06
93.5
LLM-as-a-Judge
Target Model=GPT-OSS-20B
2026.06
90
MMS probe
Target Model=GEMMA4-26...
2026.06
89.6
MMS probe
Target Model=GPT-OSS-20B
2026.06
86.5
LAT probe
Target Model=GPT-OSS-20B
2026.06
77.5
LAT probe
Target Model=GEMMA4-26...
2026.06
58.5
Feedback
Search any
task
Search any
task