Alibaba
Qwen3.8 27B (xhigh)
Built on the architectural foundation of Qwen3.5, Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks. Qwen3.8-27B brings these advances to a compact, deployment-friendly dense model: a native vision-language model that understands images and videos, with flexible thinking control, designed to carry complex, multi-step tasks through to completion with greater reliability.
Qwen model releasesModel Snapshot
TypeLanguage
ReleaseAug 14, 2026
Context1,000,000
Tagsreasoning, tool-use, vision, implicit-caching
Overall
Rank #91 across 687
33.7
Coding
Rank #76 across 330
68.1
Blended Price
Rank #233 across 432
$1.13/M
Strength Profile
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Benchmark Percentiles
Higher bars mean stronger relative placement.
Model Details
Runtime
Output Speed46.1 tok/s
First Token1.27s
Max Output131,072
Pricing
Blended$1.13/M
Input$0.500/M
All Benchmarks
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 33.7 | #91 |
| Artificial Analysis Coding Index | coding | 68.1 | #76 |
| GPQA | reasoning | 90.5% | #63 |
| Humanity's Last Exam | reasoning | 33.9% | #106 |
| SciCode |