CoWorkBench
We show this table for reference; we do not rank on it.
Qwen's internal benchmark for long-horizon professional work across computer science, finance, law, medicine, and other productivity domains.
Benchmark score on CoWorkBench — November 13, 2026
We compile the CoWorkBench rows from provider self-reports. Qwen3.8-Omni-Flash leads the table at 75.3%, followed by Qwen3.8 Max (74.8%) and Qwen3.8-Flash-Next (73.9%). We do not use these results to rank models overall.
Qwen3.8-Omni-Flash
Alibaba
Reported by AlibabaQwen3.8 Max
Alibaba
Reported by AlibabaQwen3.8-Flash-Next
Alibaba
Reported by Alibaba4 modelsAgenticCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (4 models)
All 4 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Qwen3.8-Omni-FlashAlibabaReported by Alibaba | 75.3% | Reported by Alibaba | Not reported | Closed |
| 2 | Qwen3.8 MaxAlibabaReported by Alibaba | 74.8% | Reported by Alibaba | Not reported | Open |
| 3 | Qwen3.8-Flash-NextAlibabaReported by Alibaba | 73.9% | Reported by Alibaba | Not reported | Open |
| 4 | Qwen3.8-27BAlibabaReported by Alibaba | 70.7% | Reported by Alibaba | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported CoWorkBench rows, Qwen3.8-Omni-Flash is first at 75.3%. The third row is 1.4 points behind. The broader top-10 range is 4.6 points, so many of the published results sit in a relatively narrow band.
4 models have been evaluated on CoWorkBench. The benchmark falls in the Agentic category. CoWorkBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About CoWorkBench
Year
2026
Tasks
Long-horizon professional workflows
Format
Provider-run agent score
Difficulty
Cross-domain professional work
Qwen publishes CoWorkBench as an in-house evaluation of professional cowork tasks. We store the exact provider-reported value as display-only evidence and do not treat it as independently reproduced.
Freshness and provenance
Version
CoWorkBench 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does CoWorkBench measure?
Qwen's internal benchmark for long-horizon professional work across computer science, finance, law, medicine, and other productivity domains.
Which model scores highest on CoWorkBench?
Qwen3.8-Omni-Flash by Alibaba currently leads with a score of 75.3% on CoWorkBench.
How many models are evaluated on CoWorkBench?
4 AI models have published results on CoWorkBench in the BenchLM catalog.
Compare top models on CoWorkBench
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.