Long ContextActive
Trust Score:
93

GDP.pdf

Professional-work benchmark over long PDF documents, evaluated by Artificial Analysis.

Launched: —Refresh: static
Status Assessment (active):

Actively evaluated by Artificial Analysis on current frontier models; no saturation evidence recorded.

Performance Timeline

Longitudinal progression of model scores against human baselines.

Performance & Historical Trajectory

Empirical score progression across model release dates and evaluation rounds.

Independent Vendor-Reported
ModelScoreDateSource TypeProvenance
A.X-K26.4%2026-11-15independentSource ↗
Qwen3.7 Plus12.2%2026-11-15independentSource ↗
Ling 3.1 Flash10%2026-11-15independentSource ↗
NVIDIA Nemotron 3 Nano 4B0.4%2026-11-15independentSource ↗
K2 Horizon 0.9B0.8%2026-11-15independentSource ↗
Qwen3.8-Flash-Next15.6%2026-11-15independentSource ↗
Ministral 3 8B0.8%2026-11-15independentSource ↗
Llama 4 Maverick0.8%2026-11-15independentSource ↗
Qwen3.8 2.4T A95B15.6%2026-11-15independentSource ↗
Kimi K2.7 Code11.2%2026-11-15independentSource ↗
Gemini 4 Hydrogen27.6%2026-11-15independentSource ↗
DeepSeek V4-Pro11.4%2026-11-15independentSource ↗
Llama 4 Scout0%2026-11-15independentSource ↗
Celeris-11.2%2026-11-15independentSource ↗
Ling 3.0 Tiny0.8%2026-11-15independentSource ↗
Nemotron 3.5 Bolt 74B-A9B2.8%2026-11-15independentSource ↗
Muse Ember22.8%2026-11-15independentSource ↗
DeepSeek V4 Flash Vision12%2026-11-15independentSource ↗
Mercury 2.53.6%2026-11-15independentSource ↗
Gemini 3.8 Flash21%2026-11-15independentSource ↗
Inkling Small11.2%2026-11-15independentSource ↗
Ministral 3 14B1.2%2026-11-15independentSource ↗
Ling-3.0-flash-Fin5.4%2026-11-15independentSource ↗
Beam 1.011.6%2026-11-15independentSource ↗
GLM 5.311.2%2026-11-15independentSource ↗
Qwen3 Coder Next2.6%2026-11-15independentSource ↗
Quasar 438B5.4%2026-11-15independentSource ↗
Ring-2.6-1T3.8%2026-11-15independentSource ↗
Muse Wonder Full32.6%2026-11-15independentSource ↗
Inkling 2 Small25.2%2026-11-15independentSource ↗
Nemotron 3.5 Super VL8%2026-11-15independentSource ↗
Mistral Medium 3.52.8%2026-11-15independentSource ↗
Agnes 2.5 Pro Theta14.2%2026-11-15independentSource ↗
MiniMax M3.1 Flash20.8%2026-11-15independentSource ↗
Hy421.6%2026-11-15independentSource ↗
Gemma 4 31B6%2026-11-15independentSource ↗
Ministral 3 3B0%2026-11-15independentSource ↗
gpt-oss-120b4.2%2026-11-15independentSource ↗
GPT-5.5 Instant20.2%2026-11-15independentSource ↗
Gemini 3.1 Pro17.8%2026-11-15independentSource ↗
K2 Horizon 375B A23B7.4%2026-11-15independentSource ↗
Trinity Large Thinking1.2%2026-11-15independentSource ↗
JT-4.1 Flash 236B A21B6.6%2026-11-15independentSource ↗
Hy311.2%2026-11-15independentSource ↗
Qwen3.8 Max (0902)22.8%2026-11-15independentSource ↗
Motif 35.2%2026-11-15independentSource ↗
Granite 4.2 8B0.6%2026-11-15independentSource ↗
Granite 4.2 30B1.8%2026-11-15independentSource ↗
Grok 4.720%2026-11-15independentSource ↗
Claude Haiku 5.520.8%2026-11-15independentSource ↗
GPT-6 Astra31%2026-11-15independentSource ↗
North Mini Code1.8%2026-11-15independentSource ↗
GPT-6 Luna22.8%2026-11-15independentSource ↗
GLM 5.3 Flash15.4%2026-11-15independentSource ↗
K-EXAONE 2.04.2%2026-11-15independentSource ↗
Solar Pro 47%2026-11-15independentSource ↗
Mistral Small 40%2026-11-15independentSource ↗
MiMo-V2.6-Flash9%2026-11-15independentSource ↗
K2 Horizon 7B3.2%2026-11-15independentSource ↗
MiMo-V2.6-Pro19.2%2026-11-15independentSource ↗
Qwen3.5 397B A17B12.2%2026-11-15independentSource ↗
Inkling12.8%2026-11-15independentSource ↗
Gemini 4 Carbon28.4%2026-11-15independentSource ↗
GPT-6.1 Terra28.6%2026-11-15independentSource ↗
Claude Fable 5.224.2%2026-11-15independentSource ↗
Granite 4.2 3B0.2%2026-11-15independentSource ↗
Step 514.8%2026-11-15independentSource ↗
Gemini 3.5 Flash-Lite13.6%2026-11-15independentSource ↗
GPT-6.1 Sol31%2026-11-15independentSource ↗
Apodex 1.111.2%2026-11-15independentSource ↗
Ling-3.0-flash-VL8.4%2026-11-15independentSource ↗
Kimi K3.1 Pro31.4%2026-11-15independentSource ↗
MiniCPM5-2B1.4%2026-11-15independentSource ↗
Solar Mini 45%2026-11-15independentSource ↗
LongCat 2.06.4%2026-11-15independentSource ↗
Nemotron 3 Ultra5%2026-11-15independentSource ↗
Muse Glimmer10%2026-11-15independentSource ↗
DeepSeek V4.1 Flash 101717.2%2026-11-15independentSource ↗
K2 Horizon MoVA 36B A4B4.2%2026-11-15independentSource ↗
Ling 3.1 Flash 101912.6%2026-11-15independentSource ↗
Nanbeige4.3-33B11.2%2026-11-15independentSource ↗
Gemini 4 Argon21.8%2026-11-15independentSource ↗
Claude Sonnet 5.525.8%2026-11-15independentSource ↗
Nex-N2-Pro12.6%2026-11-15independentSource ↗
Muse Wonder30.6%2026-11-15independentSource ↗
Muse Spark Open18.6%2026-11-15independentSource ↗
Qwen3.8 27B16.4%2026-11-15independentSource ↗
K2 Horizon 3.7B3.2%2026-11-15independentSource ↗
Claude Opus 5.526.2%2026-11-15independentSource ↗
Command A+1%2026-11-15independentSource ↗
Nemotron 3.5 Lightning3%2026-11-15independentSource ↗
Mistral Large 417.6%2026-11-15independentSource ↗
MiniMax M39.8%2026-11-15independentSource ↗

Human Baseline & Difficulty Horizon

Calibrated human reference points, specialist benchmarks, and ceiling thresholds.
Measured Human Scoren/aNot established
Baseline Protocol & Interpretation

No peer-reviewed human baseline has been published for this benchmark.

Metric & Scoring Methodology

Verification protocols, aggregation formulas, and specialized metric variants.
Primary Metric:accuracy (%)
Scoring Engine:llm-judge

Dataset & Compute Cost

Evaluation volume, public availability, API pricing, and local hardware requirements.
Total Dataset Size—Annotated evaluation items
Public Test SetPublicOpenly mirrored on repositories
Access GatingOpen AccessUnrestricted download
Evaluation LicenseOpen AccessDataset usage and redistribution terms
Frontier API Compute Cost:

$5 – $20 USD for full benchmark evaluation run on frontier APIs.

Recommended Local GPU Setup:

1x NVIDIA RTX 4090 (24GB) or A100 (40GB/80GB) via vLLM / SGLang

How to Run & Reproduce

Standardized evaluation protocols, CLI commands, and reproducible runner templates.
Prompt Regimezero-shot
Reasoning Modedirect
Sampling Temp0
Pass@k Budgetk = 1
Tools & SandboxPure Text
Scoring Verifierllm-judge
Option AEleutherAI LM-Evaluation-Harness (Open-Weight Models)
lm_eval --model hf --model_args pretrained=<model_path> --tasks gdp-pdf --batch_size auto
Option BOpenCompass Evaluation Framework
opencompass --datasets gdp-pdf --models <model_config>
Python APIDeterministic Inference Loop Snippet
# Standard API Evaluation Loop
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": prompt}],
    temperature=0.0,
)
Standardized Reporting Requirement:

When publishing results for GDP.pdf, always report the exact prompt template, few-shot exemplar ordering, sampling temperature (temperature=0), maximum reasoning budget tokens, and the precise timestamped model snapshot ID.

Contamination & Memorization Analysis

Audit of pretraining exposure risks, memorization vectors, and refresh policies.
Overall Contamination Risk:MEDIUM
Refresh Cadence:

Static fixed snapshot

Test Set Exposure:

Public on web / HuggingFace