Benchmarks / Cyber

UGI

A private-question suite combining sensitive-topic knowledge with willingness to follow controversial instructions. Maintained by DontPlanToEnd. Score: UGI 🏆.

166models

–Open Index weight

–last updated

benchmarkheavenpublisher

#	Model	Effort	Score	Cost / task	Harness
1	Grok 4.20	xhigh	70	–	–
2	Grok 4.0709	–	68	–	–
3	Grok 3	none	63	–	–
4	Grok 4.5	max	62	–	–
5	Grok 4.3	xhigh	59	–	–
6	Mistral Large 2411	–	59	–	–
7	DeepSeek V4 Flash	max	59	–	–
8	DeepSeek V3	none	57	–	–
9	Mistral Large 3	none	57	–	–
10	Kimi K2	medium	57	–	–
11	Mistral Medium 3	none	56	–	–
12	DeepSeek V3.0324	–	55	–	–
13	Opus 5.5	max	54	–	–
14	DeepSeek V3.2	high	54	–	–
15	Opus 4.7	max	54	–	–
16	GPT-6.1 Sol	max	53	–	–
17	o3	high	52	–	–
18	Mistral Large 2407	–	52	–	–
19	Sonnet 4.6	max	52	–	–
20	Opus 4.6	max	52	–	–
21	DeepSeek V3.1	none	51	–	–
22	DeepSeek R1	xhigh	51	–	–
23	Opus 4.8	max	51	–	–
24	GPT-5.5	xhigh	50	–	–
25	Mistral Medium 3.5	high	50	–	–
26	Llama 7 0b	–	49	–	–
27	GLM 4.7	none	49	–	–
28	Gemini 3 Flash	high	49	–	–
29	GPT-6 Astra	max	49	–	–
30	Mistral Medium 3.1	none	49	–	–
31	DeepSeek V4 Pro	max	49	–	–
32	GPT-6 Sol	max	48	–	–
33	Gemini 2.5 Pro	high	48	–	–
34	GPT-5.6 Terra	max	48	–	–
35	GPT-4.1	high	48	–	–
36	Grok 4.1 Fast	high	47	–	–
37	Gemini 3.6 Flash	xhigh	47	–	–
38	Llama 3.1 405b	none	47	–	–
39	Gemini 2.5 Flash	medium	47	–	–
40	GPT-5.4	xhigh	47	–	–
41	Qwen 2.5 32B	none	47	–	–
42	Gemini 3.8 Flash	max	47	–	–
43	Gemini 3.5 Flash	xhigh	46	–	–
44	Gemini 3.7 Flash	max	46	–	–
45	Fable 5	max	46	–	–
46	Grok 4 Fast	none	46	–	–
47	GPT-4o	high	46	–	–
48	Devstral Small 2	none	45	–	–
49	Hy3	xhigh	45	–	–
50	GPT-5.6 Sol	max	45	–	–
51	Magistral Small 2509	–	44	–	–
52	Fable 5.1	max	44	–	–
53	Qwen 2.5 72B	none	44	–	–
54	Kimi K2.5	max	44	–	–
55	Qwen 3 Max	max	44	–	–
56	DeepSeek R1.0528	–	43	–	–
57	Opus 4.5	high	43	–	–
58	MiMo V2.5 Pro	xhigh	43	–	–
59	GLM 4.6	none	43	–	–
60	o1	high	42	–	–
61	GPT-5	high	42	–	–
62	GPT-5.2025	high	42	–	–
63	Llama 4 Maverick	medium	41	–	–
64	Mistral Small 3.1	none	41	–	–
65	Llama 33 70b	–	41	–	–
66	Mistral Small 3.2	none	40	–	–
67	Llama 33	–	40	–	–
68	GLM 5.1	max	40	–	–
69	Gemma 2 7B	–	40	–	–
70	Mistral Small	none	39	–	–
71	Qwen 2.5 14B	–	39	–	–
72	Gemini 3.1 Pro	high	38	–	–
73	Llama 3.3 70b	none	38	–	–
74	Gemma 2 9B	–	38	–	–
75	Gemini 3 Pro	high	38	–	–
76	Mistral Small 3	none	37	–	–
77	Llama 3.3	none	37	–	–
78	Llama 3	–	37	–	–
79	GLM 4.5	xhigh	37	–	–
80	Gemma 1 2B	–	37	–	–
81	Kimi K2.6	max	36	–	–
82	Qwen 3	high	36	–	–
83	o4 mini	high	36	–	–
84	Qwen 3 235B-A22B	xhigh	35	–	–
85	Qwen 3 4B	none	35	–	–
86	GLM 4.6 V	low	35	–	–
87	MiMo V2.5	xhigh	35	–	–
88	Mistral Small 2409	–	35	–	–
89	MiMo V2 Flash	high	34	–	–
90	Grok 4.7	max	34	–	–
91	Opus 4	medium	34	–	–
92	GLM 5	high	34	–	–
93	Qwen 3.1	none	34	–	–
94	Mistral Small 4	high	34	–	–
95	Mistral	none	33	–	–
96	GLM 4	–	33	–	–
97	Qwen 3 30B-A3B	none	33	–	–
98	Sonnet 5	max	33	–	–
99	Gemma 3 12B	none	32	–	–
100	Qwen 3 8B	none	32	–	–
101	Qwen 3 14B	none	31	–	–
102	Magistral 2509	–	31	–	–
103	GPT-5.6 Luna	max	31	–	–
104	Llama 4 Scout	medium	31	–	–
105	Llama 3.1	max	30	–	–
106	GPT-5.3	xhigh	30	–	–
107	Llama 3 8b	none	30	–	–
108	Sonnet 4.5	high	29	–	–
109	GLM 5.2	max	29	–	–
110	GPT-5.2	xhigh	29	–	–
111	Qwen 3.5 397B-A17B	max	29	–	–
112	Qwen 3.6 27B	max	27	–	–
113	MiniMax M2	–	27	–	–
114	Llama 3.3 8b	–	25	–	–
115	Qwen 3 32B	none	25	–	–
116	Gemma 3 4B	none	25	–	–
117	Qwen 2.5 7B	–	24	–	–
118	Qwen 3.5 35B-A3B	low	23	–	–
119	Gemma 4 31B IT	high	23	–	–
120	Qwen 3.6 Plus	max	23	–	–
121	Llama 10 9b	–	22	–	–
122	GLM 4.7 Flash	none	21	–	–
123	Gemma 3 12B IT	–	21	–	–
124	Qwen 3.0	–	21	–	–
125	Llama 3.1 70b	none	21	–	–
126	Gemma 4 26B	xhigh	21	–	–
127	Gemma 3 27B	none	20	–	–
128	Gemma 3 27B IT	medium	20	–	–
129	Gemma 3 1B	none	20	–	–
130	MiniMax M2.1	–	20	–	–
131	Grok 4.6	max	20	–	–
132	gpt-oss-120b	high	20	–	–
133	MiniMax M2.5	high	18	–	–
134	MiniMax M2.7	high	18	–	–
135	Qwen 3.5 122B-A10B	none	18	–	–
136	Qwen 3.5 9B	low	17	–	–
137	Gemma 3 4B IT	–	17	–	–
138	Qwen 3.5	none	17	–	–
139	GPT-5.1	high	16	–	–
140	GLM 4.5 Air	–	16	–	–
141	Opus 4.1	high	16	–	–
142	Qwen 3.6 35B-A3B	max	16	–	–
143	Qwen 3.5 27B	high	16	–	–
144	Qwen 3 Coder	medium	15	–	–
145	Qwen 3.5 4B	low	15	–	–
146	Qwen 3 15B-A2B	–	15	–	–
147	Gemma 4 12B IT	–	15	–	–
148	Qwen 2.5	–	15	–	–
149	Gemma 2 27B IT	–	13	–	–
150	Gemma 3 1B IT	–	13	–	–
151	Sonnet 4	high	13	–	–
152	Gemma 2 9B IT	–	12	–	–
153	Llama 3 3b	–	11	–	–
154	Qwen 2.5 Coder	none	10	–	–
155	Qwen 3 16B-A3B	–	10	–	–
156	Opus 3	none	10	–	–
157	Haiku 3	medium	9	–	–
158	Haiku 4.5	xhigh	9	–	–
159	gpt-oss-20b	high	9	–	–
160	Llama 3.2 3b	none	7	–	–
161	Llama 3.2 1b	none	7	–	–
162	Llama 3.2	–	7	–	–
163	Llama 3.1 8b	medium	6	–	–
164	Qwen 3.5 2B	low	6	–	–
165	Llama 2 70b	none	4	–	–
166	Gemma 2 2B IT	–	2	–	–