Vulnerability Prediction on CWE-Prediction benchmark leaderboard · SOTA2 Research

Vulnerability Prediction on CWE-Prediction

70.4AccuracyFoundation-Sec-8B-Reasoning

Evaluation Results

Method
Foundation-Sec-8B-Reasoning2026.01	70.4	0.004
GPT-52026.01	70.1	0.002
GPT-4.12026.01	68.2	0.004
GPT-5-Mini2026.01	66.6	0.001
GPT-OSS-120B2026.01	66.1	0.002
O3-Mini2026.01	63.2	0.001
Foundation-Sec-8B-Instruct2026.01	61.6	0.003
Llama-Primus-Nemotron-70B-Instruct2026.01	61.3	0.003
GPT-5-Nano2026.01	61.2	0.003
Llama-3.3-70B-Instruct2026.01	60.7	0.003
Phi-42026.01	55.4	0.004
Llama-Primus-Merged2026.01	55	0.002
Qwen-3-14B2026.01	53.4	0.005
GPT-OSS-20B2026.01	51.9	0.006
Llama-3.1-8B-Instruct2026.01	47.3	0.003
Qwen-3-8B2026.01	46.4	0.002
DeepHat-V1-7B2026.01	36	0.003