Frontier AI Risk Evaluation on DeepSight Frontier AI Risk 1.0 (test) benchmark leaderboard · SOTA2 Research

Frontier AI Risk Evaluation on DeepSight Frontier AI Risk 1.0 (test)

74.93Overall ScoreKimi-K2-Thinking

Evaluation Results

Method	Links
Kimi-K2-Thinking2026.02		74.93	98.08	69.25	1.11	63.91	96.67	69.25	87.5	95.33	93.23
GPT-4o2026.02		73.64	96.17	79.58	33.33	42.68	75.56	71.71	96.88	100	66.89
GPT-5.22026.02		73.14	98.72	67.42	23.33	52.4	95	69.9	78.13	99	74.38
Qwen2.5-72B-Instruct2026.02		71.26	95.85	78.5	31.11	39.69	72.78	69.79	87.5	97.67	68.48
Claude-Sonnet-4.52026.02		70.86	96.81	80.92	4.44	75.16	78.33	68.5	78.13	99.67	55.82
Llama-3.3-70B-Instruct2026.02		70.19	99.68	63.17	32.22	46.18	75.56	61.8	93.75	98	61.35
Mistral-Small-24B-Instruct2026.02		69.51	90.42	79.67	32.22	39.95	71.67	55.04	96.88	98.67	61.12
Gemini-3-Flash2026.02		69.2	98.4	83.67	27.78	35.65	76.67	46.97	78.13	94.67	80.87
DeepSeek-R12026.02		68.43	94.57	86.42	2.22	41.34	96.67	60.37	96.88	83	54.4
Gemma-3-27B-IT2026.02		66.66	88.18	76.17	30	40.75	65.56	47.62	96.88	94	60.78
InternLM3-8B2026.02		66.11	96.49	57.33	28.89	39.26	70	61.61	90.63	94	56.75
Qwen3-30B-A3B-Thinking2026.02		64.57	97.44	61.5	4.44	41.65	58.89	57.93	96.88	99.33	63.1
Doubao-Seed-1.6-flash2026.02		62.21	92.65	65.17	1.11	40.16	58.9	69.92	62.5	100	69.48
GLM-4.5-Air2026.02		62.09	85.94	71.75	11.11	45.02	58.89	68.05	56.25	100	61.76