Benchmarks / Cyber

UGI Willingness

A private-question evaluation of how far a model follows challenging instructions before refusing or deviating. Maintained by DontPlanToEnd. Score: W/10 👍.

166models

–Open Index weight

–last updated

benchmarkheavenpublisher

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