Alibaba, language model
Released Aug 14, 2026. Scored on 4 boards.
The latest published score on each board, ranked against every model we track. Each score links to its source.
| Board | Score | Rank | Record | Date | Source | Readings |
|---|---|---|---|---|---|---|
| Artificial Analysis Intelligence IndexOverall | 20Xhigh | 27 of 38 | 58Claude Opus 5.5 | Oct 28, 2026 | artificialanalysis.ai | 1 |
| LMArena textOverall | 1437style control | 42 of 48 | 1524Gemini 4 Argon | Oct 22, 2026 | huggingface.co | 1 |
| Artificial Analysis Coding IndexCoding and agents | 68 | 8 of 17 | 82Claude Fable 5.1 | Oct 28, 2026 | artificialanalysis.ai | 1 |
| Output speedSpeed | 39Xhigh | 33 of 34 | 793K2 Horizon 375B A23B | Oct 28, 2026 | artificialanalysis.ai | 1 |
The closest models on the boards they share with Qwen3.8 27B.
Stories tagged with Qwen3.8 27B or naming it in the headline, newest first.
A single temperature-0 run on Allganize's Korean RAG set scored Qwen3.8 27B at 77.3% and Opus 5.5 at 76.6%, a statistical tie. Thinking off drops Qwen about 6 points, and the author notes the results are not yet repeated across runs.
Community chart of 11 open-weight models by score versus total parameters, using Artificial Analysis aggregate scores. Muse Spark Open has 816B total and 42B active parameters; Qwen3.8-Flash-Next has 180B total and 6B active. The index is a single aggregate number from AA's own runs.