Our new X account is live! Follow @wizwand_team for updates

Tool-Using Agent Safety on MT-AgentRisk 1.0 (test)

27.14Filesystem ASR

GPT-5.2

Updated 4d ago

Evaluation Results

Method	Links
GPT-5.2 2026.02		27.14	71.43	22.14	67.14	25.71	60	22.86	71.43	60	40	25.48	66.3	-	-
Claude-4.5-Sonnet 2026.02		37.14	55.74	57.86	27.14	28.57	58.57	48.57	45.71	20	73.3	44.66	43.56	-	-
GPT-5.2 2026.02		51	40	34.29	50.71	37.14	54.29	31.43	67.14	86.67	0	39.73	50.41	55.9	-24
Qwen3-Coder 2026.02		55.71	32.86	52.14	35	37.14	47.14	80	17.14	46.67	46.67	55.07	33.97	-	-
Seed-1.6 2026.02		70	18.57	64.29	15	62.86	25.71	58.57	32.86	46.67	33.33	63.29	21.92	-	-
Gemini-3-Flash 2026.02		71.43	24.29	65.71	15.71	48.57	41.43	78.57	14.29	53.33	33.33	65.48	22.74	-	-
Seed-1.6 2026.02		78.57	5.71	70.71	7.14	70	18.57	78.57	11.43	60	0	73.15	9.59	15.6	-56.3
Deepseek-v3.2 2026.02		80	4.29	80	1.43	50	27.14	91.43	2.86	73.33	0	76.16	7.12	-	-
Claude-4.5-Sonnet 2026.02		82.86	7.14	65.71	12.14	67.14	30	87.14	5.71	26.67	60	71.78	15.43	60.7	-64.6
Gemini-3-Flash 2026.02		82.86	7.14	72.86	5.71	85.71	4.29	80	8.57	80	0	78.9	6.03	20.5	-73.5
Qwen3-Coder 2026.02		88.57	1.43	67.86	13.57	80	10	90	1.43	60	0	78.08	7.67	41.8	-77.4
Deepseek-v3.2 2026.02		88.57	0	80	0	85.71	4.29	92.86	1.43	73.33	0	84.93	1.1	11.5	-84.5