Benchmarks / GDP.pdf

Reported by GDP.pdf

GDP.pdf

Rubric score on 100 real professional tasks that need reasoning over PDFs (finance, legal, insurance, engineering, HR and more).

Last updated 1 Oct 2026

Results dated
6 Jun 2026 to 1 Oct 2026
Results
49 configurations of 32 models
Unit
% of rubric
Licence
Published with permission (Surge AI results via Epoch AI)

Professional document tasks: GPT-6 Sol

Top 15 of 49 results · % of rubric, higher is better · ≈ cannot be told apart from the leader. Choose a model to highlight it.Clear highlight

  1. 1≈ GPT-5.6 Sol (max reasoning)OpenAI 30.7%
  2. 1≈ GPT-5.6 SolOpenAI 30.7%
  3. 3 Claude Opus 5.5 (max reasoning)Anthropic 30.6%
  4. 4 Claude Fable 5Anthropic 30.0%
  5. 5 Claude Fable 5 (max reasoning)Anthropic 29.8%
  6. 6 Claude Fable 5.1 (high reasoning)Anthropic 29.6%
  7. 7 Claude Fable 5.1 (max reasoning)Anthropic 27.6%
  8. 7 Muse Spark 1.3 (extra-high reasoning)Meta 27.6%
  9. 7 Muse Spark 1.3Meta 27.6%
  10. 10 GPT-6 Sol (max reasoning)OpenAI 26.4%
  11. 11 GPT-5.5 (extra-high reasoning)OpenAI 26.0%
  12. 12 GPT-5.6 Terra (medium reasoning)OpenAI 24.7%
  13. 12 GPT-5.6 TerraOpenAI 24.7%
  14. 14 Claude Opus 4.8 (max reasoning)Anthropic 24.0%
  15. 14 Claude Opus 5 (max reasoning)Anthropic 24.0%

Full results

GDP.pdf: professional document tasks, % of rubric, higher is better
#ModelProfessional document tasks
% of rubric, higher is better		Price
$ per million tokens, in / out
1≈ GPT-5.6 Sol (max reasoning)OpenAI
30.7%
$4 / $20
1≈ GPT-5.6 SolOpenAI
30.7%
$4 / $20
3 Claude Opus 5.5 (max reasoning)Anthropic
30.6%
$4 / $20
4 Claude Fable 5Anthropic
30.0%
$10 / $50
5 Claude Fable 5 (max reasoning)Anthropic
29.8%
$10 / $50
6 Claude Fable 5.1 (high reasoning)Anthropic
29.6%
$10 / $50
7 Claude Fable 5.1 (max reasoning)Anthropic
27.6%
$10 / $50
7 Muse Spark 1.3 (extra-high reasoning)Meta
27.6%
$1.25 / $4.25
7 Muse Spark 1.3Meta
27.6%
$1.25 / $4.25
10 GPT-6 Sol (max reasoning)OpenAI
26.4%
$2 / $10
11 GPT-5.5 (extra-high reasoning)OpenAI
26.0%
$5 / $30
12 GPT-5.6 Terra (medium reasoning)OpenAI
24.7%
$2 / $12
12 GPT-5.6 TerraOpenAI
24.7%
$2 / $12
14 Claude Opus 4.8 (max reasoning)Anthropic
24.0%
$5 / $25
14 Claude Opus 5 (max reasoning)Anthropic
24.0%
$5 / $25
16 Gemini 3.7 Flash (high reasoning)Google
23.8%
$1.50 / $7.50
17 Gemini 3.8 Flash (medium reasoning)Google
23.4%
$1.50 / $7.50
18 Qwen3.8-Max (0902, extra-high reasoning)Alibaba
23.2%
$2 / $6
18 Gemini 3.8 Flash (high reasoning)Google
23.2%
$1.50 / $7.50
20 GPT-6 Luna (max reasoning)OpenAI
23.0%
$0.10 / $0.50
21 GPT-5.6 Luna (medium reasoning)OpenAI
22.7%
$0.20 / $1.20
21 GPT-5.6 LunaOpenAI
22.7%
$0.20 / $1.20
23 Gemini 3.7 Flash (medium reasoning)Google
21.8%
$1.50 / $7.50
24 Claude Opus 4.7 (max reasoning)Anthropic
21.0%
$5 / $25
25 Kimi K3 (max reasoning)Moonshot AI
19.0%
$3 / $15
26 Claude Sonnet 4.6 (max reasoning)Anthropic
18.0%
$3 / $15
27 Grok 4.6 (extra-high reasoning)xAI
17.2%
$2 / $6
28 Gemini 3.1 Pro Preview (high reasoning)Google
17.0%
$2 / $12
28 Gemini 3.1 Pro PreviewGoogle
17.0%
$2 / $12
30 Qwen3.8-Flash (extra-high reasoning)Alibaba
16.6%
$0.15 / $0.47
31 Muse Spark 1.2 (extra-high reasoning)Meta
16.0%
$1.25 / $4.25
31 Grok 4.6 (high reasoning)xAI
16.0%
$2 / $6
33 Muse Spark 1.1 (medium reasoning)Meta
15.0%
$1.25 / $4.25
33 Muse Spark 1.1Meta
15.0%
$1.25 / $4.25
35 Gemini 3.5 Flash (medium reasoning)Google
14.0%
$1.50 / $9
35 Gemini 3.5 FlashGoogle
14.0%
$1.50 / $9
35 Gemini 3.6 Flash (high reasoning)Google
14.0%
$1.50 / $7.50
35 Gemini 3.6 FlashGoogle
14.0%
$1.50 / $7.50
35 Grok 4.5 (high reasoning)xAI
14.0%
$2 / $6
35 GLM-5.3-Flash (max reasoning)Z.ai
14.0%
$0.15 / $0.50
41 Muse Spark 1.2Meta
12.0%
$1.25 / $4.25
41 Kimi K2.6Moonshot AI
12.0%
$0.95 / $4
43 Gemini 3.5 Flash-Lite (high reasoning)Google
10.0%
$0.30 / $2.50
43 Gemini 3 Flash Preview (high reasoning)Google
10.0%
$0.50 / $3
43 Gemini 3 Flash PreviewGoogle
10.0%
$0.50 / $3
46 Grok 4.3 (high reasoning)xAI
8.0%
$1.25 / $2.50
46 Grok 4.3xAI
8.0%
$1.25 / $2.50
48 Nova 2 Pro Preview (no reasoning)Amazon
2.0%
–
48 Nova 2 Pro PreviewAmazon
2.0%
–

Swipe the table sideways for more columns.

Ranks follow the score as shown, so equal numbers share a rank. Results as published by GDP.pdf; we do not re-run them.

Progress over time

Professional document tasks, plotted at each model's release date, higher is better. Each model counts once, at its best setting. Scores are today's, not launch day's; 1 model without a known release date is left off. Hover or tap a mark for its model.

Best score so far

Every model at its release date. The line moves each time a new release beats every earlier one, and each new best is named where there is room.

10%20%30% 2026AprJulOct Gemini 3 Flash PreviewClaude Sonnet 4.6Claude Opus 4.7GPT-5.5Claude Fable 5GPT-5.6 Sol

Best score by lab

Each lab's best so far, stepping whenever one of its releases beats its own earlier models. The 6 labs with the best scores today; 2 others not drawn.

OpenAIAnthropicMetaGoogleAlibabaMoonshot AI

10%20%30% 2026AprJulOct OpenAIAnthropicMetaGoogleAlibabaMoonshot AI

What it measures

Rubric score on 100 real professional tasks that need reasoning over PDFs (finance, legal, insurance, engineering, HR and more).

What it does not measure

Not your documents; 100 tasks, so small differences are noise.

Source

GDP.pdf · Methodology

Surge AI; collected by Epoch AI. Licence: Published with permission (Surge AI results via Epoch AI).