BullshitBench v2 · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Opus 4.8	none	95.0%	–	–
2	Qwen 3.8 Max	minimal	95.0%	–	–
3	Sonnet 4.6	high	91.0%	–	–
4	Opus 4.5	high	90.0%	–	–
5	Opus 4.6	high	87.0%	–	–
6	Opus 4.7	none	83.0%	–	–
7	Mistral Large 4.0	none	80.0%	–	–
8	Sonnet 5	low	80.0%	–	–
9	Sonnet 4.5	high	79.0%	–	–
10	Qwen 3.8 27B	none	79.0%	–	–
11	Qwen 3.5 397B-A17B	high	78.0%	–	–
12	Fable 5.1	low	77.0%	–	–
13	Haiku 4.5	high	77.0%	–	–
14	Kimi K3	xhigh	74.0%	–	–
15	Opus 5	low	73.0%	–	–
16	GLM 5.3	minimal	72.0%	–	–
17	Qwen 3.7 Max	none	72.0%	–	–
18	Qwen 3.6 Plus	none	72.0%	–	–
19	GPT-6 Astra	max	69.0%	–	–
20	Grok 4.20	low	67.0%	–	–
21	Gemini 3.5 Flash-Lite	minimal	66.0%	–	–
22	Gemini 4	low	66.0%	–	–
23	Grok 4.6	minimal	66.0%	–	–
24	GPT-6.1 Sol	low	65.0%	–	–
25	Kimi K2.6	none	65.0%	–	–
26	MiniMax M3	xhigh	64.0%	–	–
27	Opus 5.5	low	63.0%	–	–
28	Qwen 3 Max	high	63.0%	–	–
29	MiMo V2.5 Pro	xhigh	62.0%	–	–
30	Sonnet 5.5	low	60.0%	–	–
31	GPT-6 Luna Pro	max	59.0%	–	–
32	GPT-6 Sol	none	59.0%	–	–
33	GPT-6 Luna	none	59.0%	–	–
34	Haiku 5.5	max	57.0%	–	–
35	GPT-6 Sol Pro	max	57.0%	–	–
36	DeepSeek V4.1 Flash	max	57.0%	–	–
37	Fable 5	xhigh	56.0%	–	–
38	Grok 4.5	high	55.0%	–	–
39	GPT-5.6 Terra	max	54.0%	–	–
40	Kimi K2.5	none	52.0%	–	–
41	Grok 4.3	minimal	50.0%	–	–
42	Muse Spark 1.2	minimal	50.0%	–	–
43	Haiku 3.5	medium	50.0%	–	–
44	Sonnet 3.7	medium	49.0%	–	–
45	Nemotron 3 Ultra	none	49.0%	–	–
46	GPT-5.6 Sol	low	48.0%	–	–
47	GPT-5.4	none	48.0%	–	–
48	Gemini 3 Pro	low	48.0%	–	–
49	GPT-5.5	xhigh	47.0%	–	–
50	Sonnet 3.5	medium	46.0%	–	–
51	GPT-5.2 Codex	low	45.0%	–	–
52	GPT-5.1	medium	45.0%	–	–
53	Opus 4.1	none	43.0%	–	–
54	GPT-5.6 Luna	max	41.0%	–	–
55	GPT-5.3	medium	40.0%	–	–
56	DeepSeek V4 Flash 0731	xhigh	39.0%	–	–
57	GPT-5 Codex	medium	39.0%	–	–
58	Gemini 3.6 Flash	minimal	39.0%	–	–
59	GPT-5.2	none	38.0%	–	–
60	Gemini 3.1 Pro	low	37.0%	–	–
61	GPT-5.5 Pro	xhigh	36.0%	–	–
62	DeepSeek V4 Pro 0813	max	35.0%	–	–
63	Gemini 3.7 Flash	minimal	35.0%	–	–
64	MiMo V2.5	xhigh	35.0%	–	–
65	Opus 4	medium	34.0%	–	–
66	GPT-5.4 Mini	high	32.0%	–	–
67	GPT-5.1 Codex	medium	32.0%	–	–
68	GLM 5	high	31.0%	–	–
69	GLM 5.2	xhigh	31.0%	–	–
70	Sonnet 4	high	30.0%	–	–
71	Llama 4 Maverick	medium	29.0%	–	–
72	o3	medium	26.0%	–	–
73	Gemma 4 31B IT	high	25.0%	–	–
74	GPT-5.3 Codex	low	24.0%	–	–
75	GLM 5.1	xhigh	22.0%	–	–
76	GPT-5	medium	21.0%	–	–
77	Gemma 4 26B	xhigh	21.0%	–	–
78	Qwen 3 Coder	medium	20.0%	–	–
79	Gemini 2.5 Pro	medium	20.0%	–	–
80	Llama 4 Scout	medium	20.0%	–	–
81	Gemini 3.5 Flash	xhigh	20.0%	–	–
82	Grok 4.1 Fast	high	19.0%	–	–
83	Gemini 2.5 Flash	medium	19.0%	–	–
84	DeepSeek V4 Flash	none	18.0%	–	–
85	MiMo V2 Flash	none	16.0%	–	–
86	Hy3	none	16.0%	–	–
87	Gemini 2.0 Flash	medium	15.0%	–	–
88	DeepSeek V4 Pro	xhigh	14.0%	–	–
89	Llama 3.1 8b	medium	14.0%	–	–
90	GPT-5.4 Nano	high	14.0%	–	–
91	GPT-4.1	medium	14.0%	–	–
92	DeepSeek V3.2	high	13.0%	–	–
93	GPT-4o	medium	12.0%	–	–
94	gpt-oss-120b	low	12.0%	–	–
95	Gemini 3.1 Flash-Lite	medium	11.0%	–	–
96	Gemini 3 Flash	high	10.0%	–	–
97	Haiku 3	medium	10.0%	–	–
98	Kimi K2	medium	10.0%	–	–
99	MiniMax M2.5	low	9.0%	–	–
100	MiniMax M2.7	high	8.0%	–	–
101	DeepSeek R1	xhigh	8.0%	–	–
102	GLM 4.5	xhigh	8.0%	–	–
103	o4 mini	low	8.0%	–	–
104	Mistral Small 2603	high	6.0%	–	–
105	Qwen 3 235B-A22B	none	6.0%	–	–
106	Gemma 3 27B IT	medium	3.0%	–	–
107	Mistral Large 2512	medium	2.0%	–	–
108	GPT-4o mini	medium	2.0%	–	–