DeepSeek Chinese white-collar tasks evaluation
DeepSeek-V4-Pro 53% Only in system cards Agents complete 30 Chinese professional analysis, document generation and editing tasks across 13 industries. Blind human annotators compare outputs against a baseline model.
In-house DeepSeek evaluation. The prompts are not public.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Ranked by votes People pick the better of two answers, as in an arena, and the votes are turned into ratings.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
DeepSeek-V4-Pro leads with 53%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | DeepSeek-V4-Pro | 53% | Apr 2026 | Model card |
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
One system card reports it
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 26 Apr 2026 | DeepSeek-V4-Pro technical report | 53 | Chinese white-collar tasks, overall win rate vs Opus-4.6-Max, tie 10.0, lose 37.0, non-loss 63 (Figure 11). Figure 12 dimension scores are printed but not labeled by model | Capabilities |
More in Economic work
- JobBench In use 130 tasks from 35 occupations workers most want delegated. Agent works through messy reference files, graded by fact-anchored binary rubrics. GLM-5.3 61.4%
- GDPval In use Real work tasks from 44 occupations with reference files. Experts blindly compare model deliverables with expert work. Metric is win-or-tie rate.
Seed2.1 Pro 87.9% - GDPval-AA v2.1 In use Agents with shell and web access complete 220 GDPval gold tasks from 44 occupations. Blind pairwise comparisons give Elo ratings. Claude Opus 5.5 1,846
- AA-Briefcase v1.1 In use Version 1.1 of Artificial Analysis's multi-hour office work evaluation, with Elo ratings fitted by Crowd-BT. Claude Opus 5.5 1,822
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished).
Qwen3.8-Omni-Flash 75.3% - OfficeQA In use Grounded reasoning over a large corpus of U.S. Treasury Bulletin documents: locate tables and compute precise numeric answers. Claude Opus 4.7 86.3%