# GDP.pdf strict pass rate, page images plus text, Gemini judge (GDP.pdf (Gemini judge))

> Professional document understanding over real-world PDFs, graded per rubric criterion by Gemini-3.5-Flash.

Canonical page: https://benchlm.ai/benchmarks/gdppdfgeminijudge

- Category: [Multimodal & Grounded](/multimodal-grounded)
- Last updated: October 29, 2026

## About GDP.pdf (Gemini judge)

- Year: 2026
- Tasks: Professional document prompts
- Format: Strict pass rate
- Difficulty: Professional document reasoning
- Paper: [GDP.pdf](https://surgehq.ai/blog/gdp-pdf-can-100b-ai-models-master-the-documents-that-run-the-world)

Moonshot provides each PDF as page images (longest edge 1,152 pixels) interleaved with Docling-extracted text, grades with Gemini-3.5-Flash at medium effort using the official judging prompt, and reports the strict pass rate among fully scored responses, averaged over five runs. The judge and input format differ from Anthropic's and Artificial Analysis's GDP.pdf rows, so BenchLM keeps this protocol on its own key.

GDP.pdf (Gemini judge) is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (1 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Kimi K3.1 Pro](/models/kimi-3-1-pro) | Moonshot AI | 34.6% |

## FAQ

### What does GDP.pdf (Gemini judge) measure?

Professional document understanding over real-world PDFs, graded per rubric criterion by Gemini-3.5-Flash.

### Which model scores highest on GDP.pdf (Gemini judge)?

Kimi K3.1 Pro by Moonshot AI currently leads with a score of 34.6% on GDP.pdf (Gemini judge).

### How many models are evaluated on GDP.pdf (Gemini judge)?

1 AI models have published results on GDP.pdf (Gemini judge) in the BenchLM catalog.

### Does GDP.pdf (Gemini judge) affect BenchLM's overall score?

Not directly. GDP.pdf (Gemini judge) is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.