Docbench - Planted Problem Recall - Moderate: leaderboard
Docbench in Gezel: fixed document-task adapters, parser/repair/retry behavior, native checks plus Claude Opus 5.5 rubric judge. Unusable outputs score zero. 32, 768 context except Apple 8, 192. Mostly Q4 is not affirmative precision evidence for each row; preserve explicit Q2/Q4 only. CLI/API labels are run configuration, not distinct public models. Native Apple Foundation Models row retained in pinned capture but not emitted until exact checkpoint identity is established. Native planted-problem manual review cohort; severity classes are separate denominators. Real/trivial/wrong suggestions per document are diagnostics, not added as ability scores. Metric: Planted problems found (%). Source: https://bendyline.com/posts/docbench.html
Metric: Planted problems found (%). Source: bendyline.com. Saturation forecast: Around February 2027. 10 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.8 27B (Q4) | 73 | |
| 2 | Qwen 3.5 9B | 55 | #822 |
| 3 | Gemma 4 12B | 55 | #931 |
| 4 | Qwen 3.6 35B A3B | 50 | #484 |
| 5 | Gemma 4 E2B | 36 | #4291 |
| 6 | Ornith-1.5-9B | 36 | #843 |
| 7 | Qwen 3.5 4B | 18 | #1227 |
| 8 | MiniCPM5-2B | 13 | #1452 |
| 9 | Qwen 3.5 0.8B | 0 | #8136 |
| 10 | MiniCPM5-1B | 0 | #9079 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=docbench-planted-problem-recall-moderate · How It Works · Data refreshed daily, snapshot 2026-11-17.