Benchmarks / Professional

GDP.pdf

GDP.pdf: document-heavy professional tasks over PDFs graded against rubric criteria; share of tasks passing all criteria.

146models

–Open Index weight

–last updated

Artificial Analysispublisher

#	Model	Effort	Score	Cost / task	Harness
1	Muse Wonder Full (Max)	max	32.6%	–	–
2	GPT-6 Astra	xhigh	32.2%	$1.91	–
3	GPT-6.1 Sol	high	32.0%	$0.349	–
4	Muse Wonder	max	30.6%	$0.723	–
5	Opus 5.5	high	28.8%	$0.825	–
6	Fable 5.1	low	28.0%	$1.92	–
7	GPT-5.6 Sol	high	27.8%	$0.700	–
8	Muse Spark 1.3	max	26.6%	$0.205	–
9	Sonnet 5.5	max	25.8%	$0.794	–
10	GPT-6 Sol	max	25.2%	$0.425	–
11	GPT-5.6 Terra	xhigh	24.6%	$0.416	–
12	Fable 5	max	24.0%	$2.24	–
13	GPT-5.6 Luna	max	24.0%	$0.057	–
14	Gemini 3.7 Flash	high	23.6%	$0.091	–
15	Grok 4.7	high	23.2%	$0.302	–
16	Qwen 3.8 Max	max	22.8%	$0.333	–
17	Opus 4.8	max	22.8%	$1.22	–
18	Gemini 3.8 Flash	medium	22.8%	$0.087	–
19	GPT-6 Luna	max	22.8%	$0.029	–
20	GPT-5.5	high	22.6%	$0.926	–
21	Kimi K3	max	22.0%	$0.640	–
22	Gemini 4	high	21.8%	$0.259	–
23	Opus 5	max	21.6%	$1.09	–
24	Haiku 5.5	max	20.8%	$0.053	–
25	GPT-5.5 Instant	–	20.2%	$0.543	–
26	Gemini 3.5 Flash	high	19.8%	$0.188	–
27	MiMo V2.6 Pro	–	19.2%	$0.024	–
28	Grok 4.5	high	18.8%	$0.244	–
29	Muse Spark Open (Xhigh)	xhigh	18.6%	$0.188	–
30	Mistral Large 4	–	18.6%	$0.216	–
31	Grok 4.6	medium	17.8%	$0.252	–
32	Gemini 3.1 Pro	–	17.8%	$0.238	–
33	Muse Spark 1.2	xhigh	17.4%	$0.191	–
34	Gemini 3.6 Flash	high	17.4%	$0.090	–
35	Qwen 3.8 27B	xhigh	16.4%	$0.096	–
36	Sonnet 4.6	max	15.8%	$0.606	–
37	Qwen 3.8 2.4T-A95B	–	15.6%	$0.190	–
38	Qwen 3.8 Flash	–	15.6%	$0.022	–
39	GLM 5.3 Flash	max	15.4%	$0.032	–
40	Step 5 Preview	–	14.8%	$0.122	–
41	Muse Spark 1.1	xhigh	14.4%	$0.185	–
42	Gemini 3.5 Flash-Lite	high	13.6%	$0.039	–
43	DeepSeek V4 Pro 0424	max	13.4%	$0.029	–
44	GPT-5.4 Mini	xhigh	13.4%	$0.237	–
45	Sonnet 5	max	13.2%	$0.615	–
46	Kimi K2.6	–	13.0%	$0.152	–
47	DeepSeek V4.1 Flash	max	12.8%	$0.041	–
48	Inkling	xhigh	12.8%	–	–
49	Nex-N2-Pro	–	12.6%	–	–
50	Qwen 3.7 Plus	–	12.2%	$0.064	–
51	Qwen 3.5 397B-A17B	–	12.2%	$0.078	–
52	DeepSeek V4 Flash	max	12.0%	$0.044	–
53	Beam 1.0	–	11.6%	–	–
54	DeepSeek V4 Pro 0813	max	11.4%	$0.110	–
55	GLM 5.3	max	11.2%	$0.173	–
56	Apodex 1.1	–	11.2%	$0.054	–
57	Kimi K2.7 Code	–	11.2%	$0.140	–
58	Inkling Small	–	11.2%	$0.037	–
59	Hy3	–	11.2%	$0.0090	–
60	DeepSeek V4 Flash 0731	max	11.0%	$0.047	–
61	Qwen 3.6 27B	–	11.0%	$0.090	–
62	DeepSeek V4 Flash 0420	max	10.8%	$0.0078	–
63	GLM 5.2	max	10.4%	$0.099	–
64	Gemini 2.5 Pro	–	10.2%	$0.105	–
65	Ling 3.1 Flash	–	10.0%	$0.028	–
66	Muse Glimmer (High)	high	10.0%	$0.023	–
67	MiniMax M3	–	9.8%	$0.053	–
68	MiMo V2.6 Flash	–	9.0%	$0.020	–
69	JT-4.1 Flash 236B A21B (Reasoning)	–	8.6%	–	–
70	Qwen 3.7 Max	max	8.6%	$0.132	–
71	GLM 5.1	–	8.4%	$0.052	–
72	Ling 3.0 Flash	–	8.4%	–	–
73	GPT-5 Mini	high	8.4%	$0.036	–
74	Nemotron 3.5 Super VL	–	8.0%	$0.021	–
75	GPT-5.4 Nano	xhigh	7.8%	$0.057	–
76	Gemini 3.1 Flash-Lite	high	7.8%	$0.027	–
77	K2 Horizon 375B A23B	–	7.4%	–	–
78	Solar Pro 4	–	7.0%	$0.022	–
79	Qwen 3.6 35B-A3B	–	6.6%	$0.044	–
80	Qwen 3.5 9B	–	6.6%	$0.014	–
81	A.X-K2	–	6.4%	–	–
82	LongCat 2.0	–	6.4%	$0.057	–
83	MiniMax M2.7	–	6.0%	–	–
84	Gemma 4 31B	–	6.0%	$0.0000	–
85	Grok 4.3	high	5.8%	$0.050	–
86	Quasar 438B (Max, Based on GLM-5.2)	max	5.4%	$0.054	–
87	Ling 3.0 Flash Fin	–	5.4%	–	–
88	Ling 3.0 Flash	–	5.4%	–	–
89	DeepSeek V3.1	–	5.4%	–	–
90	Motif 3	–	5.2%	–	–
91	Sonnet 4.5	–	5.2%	$0.233	–
92	Solar Mini 4	–	5.0%	$0.011	–
93	Nemotron 3 Ultra	max	5.0%	$0.031	–
94	Qwen 3 235B-A22B	–	4.8%	–	–
95	K2 Horizon MoVA 36B A4B	–	4.2%	–	–
96	MiMo V2.5 Pro	–	4.0%	$0.019	–
97	MiMo V2.5	–	4.0%	$0.0061	–
98	gpt-oss-120b	high	4.0%	$0.0092	–
99	Haiku 4.5	–	3.8%	$0.109	–
100	Ring-2.6-1T	xhigh	3.8%	$0.024	–
101	Mercury 2	high	3.8%	$0.011	–
102	Mercury 2.5	–	3.6%	$0.012	–
103	Qwen 3.5 122B-A10B	–	3.4%	$0.040	–
104	K2 Horizon 7B	–	3.2%	–	–
105	K2 Horizon 3.7B	–	3.2%	–	–
106	Nemotron 3.5 Lightning	–	3.0%	$0.0050	–
107	Nemotron 3.5 Bolt 74B-A9B	–	2.8%	–	–
108	Mistral Medium 3.5	high	2.8%	$0.139	–
109	Nemotron 3 Super 120B A12B (Reasoning)	–	2.6%	$0.011	–
110	Qwen 3 Coder	none	2.6%	$0.022	–
111	DeepSeek R1	–	2.4%	$0.077	–
112	Devstral 2	none	2.4%	–	–
113	o3 mini	high	2.0%	–	–
114	gpt-oss-20b	high	2.0%	$0.0041	–
115	Granite 4.2 30B	–	1.8%	$0.0064	–
116	North Mini Code	–	1.8%	$0.0000	–
117	DeepSeek V3	none	1.6%	$0.0086	–
118	Solar Pro 3	high	1.6%	$0.010	–
119	Qwen 3 32B	–	1.6%	–	–
120	MiniCPM5-2B	–	1.4%	–	–
121	Qwen 3 8B	–	1.4%	–	–
122	Trinity Large Thinking	–	1.2%	$0.011	–
123	Mistral Large 3	none	1.2%	$0.029	–
124	Mistral Medium 3.1	none	1.2%	–	–
125	Qwen 3 14B	–	1.2%	–	–
126	Celeris-1	none	1.2%	$0.0075	–
127	Ministral 3 14B	none	1.2%	$0.011	–
128	Command A	–	1.0%	$0.0000	–
129	Gemma 4 E4B (Reasoning)	–	1.0%	$0.0009	–
130	NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)	–	1.0%	$0.0032	–
131	Devstral Small 2	none	1.0%	–	–
132	Mistral Small 3.1	none	1.0%	$0.0042	–
133	Gemma 3 12B	none	1.0%	$0.0000	–
134	Ling 3.0 Tiny	–	0.8%	$0.0000	–
135	Llama 4 Maverick	none	0.8%	$0.0003	–
136	Ministral 3 8B	none	0.8%	$0.0084	–
137	K2 Horizon 0.9B	–	0.8%	–	–
138	Granite 4.2 8B	–	0.6%	$0.0030	–
139	Qwen 3 30B-A3B	–	0.4%	$0.017	–
140	NVIDIA Nemotron 3 Nano 4B	–	0.4%	–	–
141	Granite 4.2 3B	–	0.2%	$0.0014	–
142	Gemma 3 27B	none	0.2%	$0.0000	–
143	Mistral Small 4	–	0.0%	$0.0026	–
144	Mistral Small 3.2	none	0.0%	$0.0016	–
145	Llama 4 Scout	none	0.0%	$0.0003	–
146	Ministral 3 3B	none	0.0%	$0.0025	–