Cybersecurity Knowledge Evaluation on Cybermetric 2000

94.1Accuracy

GPT-5

Updated 23d ago

Evaluation Results

Method
GPT-5 2026.01	94.1	0.002
GPT-4.1 2026.01	93.7	0.003
GPT-5-Mini 2026.01	93.2	0.002
Llama-3.3-70B-Instruct 2026.01	93	0.003
O3-Mini 2026.01	93	0.003
GPT-OSS-120B 2026.01	92.6	0.002
gpt-oss-120B 2025.10	92.55	-
GPT-5-Nano 2026.01	91.8	0.003
Phi-4 2026.01	91.2	0.004
Qwen-3-14B 2026.01	91	0.002
Llama-Primus-Nemotron-70B-Instruct 2026.01	90.6	0.003
Qwen-3-8B 2026.01	90.3	0.002
gpt-oss-20B 2025.10	90.2	-
CyberPal-2.0-14B 2025.10	89.95	-
CyberPal-2.0-8B 2025.10	89.85	-
Qwen3-14B 2025.10	89.85	-
GPT-OSS-20B 2026.01	89.3	0.005
CyberPal-2.0-20B 2025.10	89.05	-
Qwen3-8B 2025.10	88.45	-
CyberPal-2.0-4B 2025.10	87.8	-
Qwen3-4B 2025.10	87.4	-
Llama-Primus-Merged 2026.01	86.2	0.002
Llama-3.1-8B-Instruct 2026.01	85.1	0.004
Foundation-Sec-8B-Instruct 2026.01	84.7	0.005
Foundation-Sec-8B-Reasoning 2026.01	84.3	0.003
DeepHat-V1-7B 2026.01	73	0.005