Doubao Work tops SuperCLUE-Office October benchmark at 85.33, ahead of Kimi Work (84.57)

ByteDance's Doubao Work ranked first in SuperCLUE's October 2026 SuperCLUE-Office end-to-end office-agent benchmark with an overall score of 85.33, according to results published on SuperCLUE's site and reported by IT Home and Pandaily. Moonshot AI's Kimi Work placed second at 84.57 and Tencent's WorkBuddy third at 83.99. The scores are the organizer's own; they are not an independent re-test.

What the benchmark measures

SuperCLUE-Office tests whether office agents can finish real deliverables in Word, PowerPoint, Excel and charts. The October run covered three dimensions: document processing, materials drafting and data analysis. These spanned 13 secondary tasks.

Scoring covered five capability areas:

• File and environment operations
• Multi-file planning
• Constraint following
• Content quality
• Safety and permission boundaries

Eight Chinese office agents took part: Doubao Work, Kimi Work, WorkBuddy, Baidu Dazi, Alibaba's QwenWork, SenseTime's Raccoon Work, iFLYTEK Spark Work and 360's Nano Work.

Overall scores

Rank	Agent	Score
1	Doubao Work	85.33
2	Kimi Work	84.57
3	WorkBuddy	83.99
4	Baidu Dazi	82.68
5	QwenWork	78.38
6	Raccoon Work	69.20
7	iFLYTEK Spark Work	69.12
8	Nano Work	53.13

The top four finished within 2.65 points of one another. SuperCLUE said the gap between the highest and lowest overall scores was 32.20 points, showing a clear gradient below the leading pack.

Category results

Document processing: Doubao Work led at 87.12, nearly five points ahead of the next entry. Baidu Dazi and WorkBuddy followed closely. All eight products scored above 72 in this dimension.

Data analysis: Kimi Work led at 88.00, followed by WorkBuddy at 86.13 and Doubao Work at 81.50. SuperCLUE said this category had the widest spread, with a 72-point gap between the best and worst scores and an average of 71.72.

Materials drafting: SuperCLUE described the top of this table as tighter but the lower ranks as uneven. The category had a 47.75-point max gap and a 73.62 average.

SuperCLUE's assessment and caveats

SuperCLUE said office agents are moving into everyday workflows but remain uneven on harder analysis and drafting jobs. It flagged those areas as the next competitive focus. Because the results come from a single organizer's harness and task set, they should be read as one public leaderboard rather than a universal ranking of every office agent.

What this means

The overall ranking hides different strengths. Doubao Work won document processing, while Kimi Work won data analysis, so the best agent depends on the workload. Teams that mostly produce documents and teams that mostly analyze data would read this table differently.

The category spreads also show where products diverge most. Every agent cleared 72 in document processing, but data analysis spanned a 72-point gap between best and worst. Overall scores in the top four are close, while scores in analysis and drafting separate the field more widely.

Finally, this is a single-organizer result built on SuperCLUE's own task set. It is a useful reference point for comparing these eight Chinese office agents, not a substitute for testing on your own documents and permission requirements.