Docbench - Writing: 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. 102 public cases, 142 outputs; some models have three repetitions. Overall family weights25/25/20/10/15/5; 95% range is uncertainty, not another observation. Retain each native family grader mixture. Metric: Native task quality score (%). Source: https://bendyline.com/posts/docbench.html

Metric: Native task quality score (%). Source: bendyline.com. Saturation forecast: Estimated already saturated. 29 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-6.1 Sol93#12
2GPT-5.6 Sol92#33
3Claude Opus 5.591#7
4Claude Haiku 4.589#580
5Gemma 4 31B89#385
6Qwen 3.6 35B A3B89#484
7Claude Sonnet 5.589#21
8Qwen 3.8 27B (Q4)89
9Claude Sonnet 4.688#173
10Gemma 4 26B A4B88#581
11Qwen 3.5 9B87#822
12Ornith-1.5-9B87#843
13Qwen 3.8 27B (Q2)87
14Gemma 4 E4B86#2142
15Qwen 3.5 4B85#1227

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=docbench-writing · How It Works · Data refreshed daily, snapshot 2026-11-17.