WorkBuddyBench-Office
Agents edit office files and reconcile records across a workspace. One task requires updating an eleven-slide presentation from a spreadsheet while preserving unrelated slides.
Tencent ranks 22 models on 50 Office tasks in CodeBuddy Code. The score combines file checks with model judgments about the deliverable.
- Made by
- Tencent
- Released
- July 2026
- Kind
- Documents and spreadsheets
- Level
- University
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Interaction
- Agentic
- Environment
- Terminal
- Items from
- Real work and use, Written by experts
- Refresh
- New versions replace it
- Run by
- Anyone
- In views
- Agents
- Strongest model tested
- Claude Opus 5.5 #5 on the capability index
- Last model added
- October 2026
- Scores refreshed
- November 20th, 2026
Kimi K3.1 Pro leads with 86.4% Kimi K3.1 Pro leads with 86.4%
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
- 1 Kimi K3.1 Pro think 86.4%
- 2 Claude Opus 5.5 think 85.2%
- 3 Hy4-dev think 84.7%
- 4
DeepSeek V4.1 Flash 1017 think 82.7%
- 5
MiniMax-M3.1-Flash think 82.4%
- 6
Tencent Hy3 think 82.1%
- 7
Tencent Hy4 preview think 81.2%
- 8 Kimi K3 think 81%
- 9 GLM-5.3 think 79.8%
- 10 GLM-5.2 think 79.6%
- 11
Step 5 Preview think 79.1%
- 12
DeepSeek-V4-Pro think 78.7%
- 13
MiniMax-M3 think 78.3%
- 14
DeepSeek-V4-Flash think 77.5%
- 15 MiMo-V2.6 Pro think 75.7%
Source: Official WorkBuddyBench-Office leaderboard
- 1 Kimi K3.1 Pro think 86.4%
- 2 Claude Opus 5.5 think 85.2%
- 3
DeepSeek V4.1 Flash 1017 think 82.7%
- 4
MiniMax-M3.1-Flash think 82.4%
- 5
Tencent Hy3 think 82.1%
- 6
Tencent Hy4 preview think 81.2%
- 7 Kimi K3 think 81%
- 8 GLM-5.3 think 79.8%
- 9 GLM-5.2 think 79.6%
- 10
Step 5 Preview think 79.1%
- 11
DeepSeek-V4-Pro think 78.7%
- 12
MiniMax-M3 think 78.3%
- 13
DeepSeek-V4-Flash think 77.5%
- 14 MiMo-V2.6 Pro think 75.7%
Source: Official WorkBuddyBench-Office leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Kimi K3.1 Pro | 86.4% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 2 | Claude Opus 5.5 | 85.2% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
| 3 | Hy4-dev open weights | 84.7% | think, CodeBuddy Code | Official leaderboard | |
| 4 | | 82.7% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 5 | | 82.4% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 6 | | 82.1% | think, CodeBuddy Code | Jul 2026 | Official leaderboard |
| 7 | | 81.2% | think, CodeBuddy Code | Aug 2026 | Official leaderboard |
| 8 | Kimi K3 | 81% | think, CodeBuddy Code | Jul 2026 | Official leaderboard |
| 9 | GLM-5.3 | 79.8% | think, CodeBuddy Code | Aug 2026 | Official leaderboard |
| 10 | GLM-5.2 | 79.6% | think, CodeBuddy Code | Jun 2026 | Official leaderboard |
| 11 | | 79.1% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
| 12 | | 78.7% | think, Claude Code | Apr 2026 | Official leaderboard |
| 13 | | 78.3% | think, CodeBuddy Code | Jun 2026 | Official leaderboard |
| 14 | | 77.5% | think, Claude Code | Apr 2026 | Official leaderboard |
| 15 | MiMo-V2.6 Pro | 75.7% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Kimi K3.1 Pro | 86.4% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 2 | Claude Opus 5.5 | 85.2% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
| 3 | | 82.7% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 4 | | 82.4% | think, CodeBuddy Code | Oct 2026 | Official leaderboard |
| 5 | | 82.1% | think, CodeBuddy Code | Jul 2026 | Official leaderboard |
| 6 | | 81.2% | think, CodeBuddy Code | Aug 2026 | Official leaderboard |
| 7 | Kimi K3 | 81% | think, CodeBuddy Code | Jul 2026 | Official leaderboard |
| 8 | GLM-5.3 | 79.8% | think, CodeBuddy Code | Aug 2026 | Official leaderboard |
| 9 | GLM-5.2 | 79.6% | think, CodeBuddy Code | Jun 2026 | Official leaderboard |
| 10 | | 79.1% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
| 11 | | 78.7% | think, Claude Code | Apr 2026 | Official leaderboard |
| 12 | | 78.3% | think, CodeBuddy Code | Jun 2026 | Official leaderboard |
| 13 | | 77.5% | think, Claude Code | Apr 2026 | Official leaderboard |
| 14 | MiMo-V2.6 Pro | 75.7% | think, CodeBuddy Code | Sep 2026 | Official leaderboard |
Rankings by part
We read 2 parts from the benchmark’s own leaderboard: Rule checks and LLM Judge. Each ranks models on one part of the test. Tencent Hy3 leads both. Arrows show how far a model moved from its overall rank among the models we track.
Rule checks
5 models we track have a score here.
- 1
Tencent Hy3 ↑4 81.2%
- 2 GLM-5.2 ↑7 79.3%
- 3
DeepSeek-V4-Pro ↑8 78.9%
- 4
MiniMax-M3 ↑8 78.3%
- 5
DeepSeek-V4-Flash ↑8 77.4%
LLM Judge
5 models we track have a score here.
- 1
Tencent Hy3 ↑4 86.3%
- 2 GLM-5.2 ↑7 81.8%
- 3
DeepSeek-V4-Pro ↑8 80.4%
- 4
MiniMax-M3 ↑8 78.4%
- 5
DeepSeek-V4-Flash ↑8 78.1%
4 models have held number one on the official leaderboard since its release in July 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official WorkBuddyBench-Office leaderboard
Claude Opus 5.5 on the official leaderboard passed 85% in September 2026
That was 2 months after its release. By the same age, 21% of the benchmarks from 2026 had passed 85%. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official WorkBuddyBench-Office leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
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
- Agents' Last Exam v1.1 In use Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. GPT-6.2 Astra 47.5%
- CoWorkBench In use Agent completes long-horizon productivity tasks across computer science, finance, law and medicine. Qwen in-house scoring (method unpublished). Claude Fable 5 75.9%
Data audit
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (15 entries).
- Pass Models scored 14 models we track, plus 1 other entry on its leaderboard.
- Pass Newest score The newest is from October 2026, for Kimi K3.1 Pro.
- Note Frontier models 1 of the top 10 has a score, which is fine for a retired test. The strongest tested is Claude Opus 5.5 (#5).
- Note Headroom Saturated. The best is 86.4%, past 85% since September 2026.
Leaderboard import 3 pass
- Pass Importer We read the board on workbuddybench.com: 19 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.