GDP.pdf

An implementation of Surge AI's GDP.pdf benchmark, testing whether language models can reason over long, real-world professional documents and satisfy detailed task-specific criteria.

Intelligence Index weight: 5%
Models: 148 current, 254 in total
Median: 13.2%
Range: 0.0%–32.6%

Score distribution

Lowest: 0.0%
Median: 13.2%
Highest: 32.6%

Leaderboard

Include deprecated models

Rank	Model	Score	Cost / task	Time / task	Output tokens / task
1	Muse Wonder Full (Max) Meta	32.6%	—	—	11.5K
2	GPT-6 Astra (Xhigh) OpenAI	32.2%	$1.91	2m 23s	5.58K
3	GPT-6.1 Sol (High) OpenAI	32.0%	$0.35	46.9s	2.67K
4	GPT-6.1 Sol (Xhigh) OpenAI	31.8%	$0.37	1m 21s	4.54K
5	Kimi K3.1 Pro Kimi	31.4%	$0.52	9m 49s	19.1K
6	GPT-6.1 Sol (Max) OpenAI	31.0%	$0.42	2m 53s	9.72K
7	GPT-6 Astra (Max) OpenAI	31.0%	$2.08	4m 6s	9.25K
8	GPT-6 Astra (High) OpenAI	31.0%	$1.79	1m 22s	3.22K
9	Muse Wonder (Max) Meta	30.6%	$0.72	1m 31s	11.1K
10	GPT-6 Astra (Low) OpenAI	30.4%	$1.70	31.5s	1.25K
11	GPT-6 Astra (Medium) OpenAI	30.4%	$1.72	40.3s	1.72K
12	GPT-6.1 Sol (Medium) OpenAI	30.0%	$0.34	25.1s	1.48K
13	Gemini 4 Carbon (Medium) Google	29.6%	$0.48	1m 28s	6.53K
14	Muse Wonder (High) Meta	29.2%	$0.60	40.7s	4.93K
15	Claude Opus 5.5 (High, Default Fallback) Anthropic	28.8%	$0.83	56.9s	5.97K
16	Gemini 4 Carbon (Low) Google	28.6%	$0.38	26.6s	1.71K
17	GPT-6.1 Terra (Max) OpenAI	28.6%	$0.24	3m 21s	14.8K
18	Gemini 4 Carbon (High) Google	28.4%	$0.85	4m 45s	23.4K
19	GPT-6.1 Terra (Medium) OpenAI	28.4%	$0.17	15.5s	1.07K
20	GPT-6.1 Terra (High) OpenAI	28.0%	$0.17	28.1s	1.96K
21	GPT-6.1 Sol (Low) OpenAI	27.0%	$0.33	20.3s	1.15K
22	Claude Fable 5.2 (Low, Default Fallback) Anthropic	26.8%	$1.53	34.0s	3.15K
23	Claude Opus 5.5 (Xhigh, Default Fallback) Anthropic	26.6%	$0.96	1m 59s	12.9K
24	GPT-6.1 Terra (Xhigh) OpenAI	26.6%	$0.18	58.4s	4.19K
25	Claude Opus 5.5 (Max, Default Fallback) Anthropic	26.2%	$1.55	6m 2s	42.3K
26	Claude Sonnet 5.5 (Max, Default Fallback) Anthropic	25.8%	$0.79	4m 48s	44.2K
27	Claude Opus 5.5 (Low, Default Fallback) Anthropic	25.6%	$0.76	26.8s	2.9K
28	Claude Opus 5.5 (Medium, Default Fallback) Anthropic	25.6%	$0.80	43.2s	4.46K
29	GPT-6.1 Terra (Low) OpenAI	25.6%	$0.16	9.0s	612
30	Claude Sonnet 5.5 (High, Default Fallback) Anthropic	25.2%	$0.41	39.4s	5.56K
31	Inkling 2 Small Thinking Machines	25.2%	$0.057	3m 8s	24K
32	Claude Sonnet 5.5 (Xhigh, Default Fallback) Anthropic	24.6%	$0.46	1m 17s	10.7K
33	Claude Fable 5.2 (Max, Default Fallback) Anthropic	24.2%	$2.52	4m 31s	27.8K
34	Claude Fable 5.2 (Medium, Default Fallback) Anthropic	23.8%	$1.56	41.7s	3.78K
35	GPT-6.1 Terra (Non-reasoning) OpenAI	23.2%	$0.16	9.1s	629
36	Grok 4.7 (High) SpaceXAI	23.2%	$0.30	3m 15s	13.1K
37	Gemini 3.8 Flash (Medium) Google	22.8%	$0.087	52.7s	5.91K
38	GPT-6 Luna (Max) OpenAI	22.8%	$0.029	3m 47s	24.8K
39	Muse Ember (Max) Meta	22.8%	$0.017	20.9s	9.52K
40	Qwen3.8 Max (0902) Alibaba	22.8%	$0.33	3m 41s	11.3K
41	Muse Ember (Xhigh) Meta	22.4%	$0.016	14.3s	6.55K
42	Claude Fable 5.2 (High, Default Fallback) Anthropic	21.8%	$1.64	1m 4s	5.96K
43	Gemini 4 Argon (High) Google	21.8%	$0.52	2m 4s	8.43K
44	Tencent Hy4 Tencent	21.6%	$0.17	—	24.8K
45	Claude Fable 5.2 (Xhigh, Default Fallback) Anthropic	21.4%	$1.85	1m 47s	11.1K
46	Grok 4.7 (Low) SpaceXAI	21.4%	$0.25	1m 1s	3.91K
47	Gemini 3.8 Flash (High) Google	21.0%	$0.11	1m 51s	12.8K
48	Claude Haiku 5.5 (Max) Anthropic	20.8%	$0.053	4m 36s	70.2K
49	MiniMax-M3.1-Flash MiniMax	20.8%	$0.036	2m 10s	20.8K
50	Claude Sonnet 5.5 (Medium, Default Fallback) Anthropic	20.2%	$0.38	20.4s	2.84K

1
Muse Wonder Full (Max)
Meta
32.6%
—
—
11.5K
2

Data from Artificial Analysis

Command Palette

GDP.pdf

Score distribution

Leaderboard