AA GDP.pdf - STEM/Research - Criterion Pass Rate: leaderboard
Artificial Analysis's implementation of Surge AI's GDP.pdf restricted to its STEM/Research domain: the mean share of task-specific atomic criteria satisfied on professional tasks over long PDF documents. Reported 0-100 (%); higher is better. A criterion pass rate, not the all-criteria pass rate shown on the AA GDP.pdf board.
Metric: Criterion Pass Rate (%). Source: artificialanalysis.ai. Saturation forecast: Around February 2027. 242 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Wonder Full (Max) | 80.46 | #12 (Muse Wonder Full) |
| 2 | Gemini 4 Carbon (High) | 80.11 | #1 (Gemini 4 Carbon) |
| 3 | GPT-6 Astra (Medium) | 79.89 | #6 (GPT-6 Astra) |
| 4 | Muse Wonder (Max) | 79.78 | #18 (Muse Wonder) |
| 5 | GPT-6 Astra (xHigh) | 79.49 | #6 (GPT-6 Astra) |
| 6 | GPT-6.1 Sol (High) | 79.37 | #7 (GPT-6.1 Sol) |
| 7 | GPT-6 Sol (High) | 79.19 | #19 (GPT-6 Sol) |
| 8 | GPT-6 Astra (Max) | 79.02 | #6 (GPT-6 Astra) |
| 9 | GPT-6 Sol (xHigh) | 79.01 | #19 (GPT-6 Sol) |
| 10 | GPT-6.1 Sol (xHigh) | 78.85 | #7 (GPT-6.1 Sol) |
| 11 | GPT-6 Astra (High) | 78.8 | #6 (GPT-6 Astra) |
| 12 | GPT-6.1 Sol (Max) | 78.6 | #7 (GPT-6.1 Sol) |
| 13 | GPT-6 Sol (Max) | 78 | #19 (GPT-6 Sol) |
| 14 | GPT-6.1 Sol (Low) | 77.94 | #7 (GPT-6.1 Sol) |
| 15 | GPT-6.1 Terra (Low) | 77.88 | #15 (GPT-6.1 Terra) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=aa-gdp-pdf-stem-research-criterion-pass-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.