HolyGrAIL

All tasks / Models

Qwen3.8-27B

Dense 27B open-weights Qwen3.8 model under Apache 2.0 that fits a 24 GB GPU at 4-bit quantization.

  • Alibaba (Qwen)
  • Open weights
  • Released Aug 2026

Official page

Where it ranks

  • A
    Chat & assistants
    Models · Local · 1,438 on LMArena Text Arena (overall, style control)
  • B
    Writing
    Models · Local · 1,366 on LMArena Text Arena: Creative Writing (style control)
  • –
    Documents & OCR
    Models · Local · no result on this task’s benchmarks yet
  • C
    Coding
    Models · Local · 1,592 on LMArena WebDev / Code Arena (overall)
  • A
    Image understanding
    Models · Local · 1,244 on LMArena Vision Arena (overall, style control)
  • C
    Workflow automation
    Models · Local · 48.2% on AutomationBench-AA (Artificial Analysis)

Facts

Vendor
Alibaba (Qwen)
Family
Qwen (small)
Released
Aug 2026
Weights
Open: download and run it yourself
License
Apache-2.0
Parameters
27B
GPU memory to run locally
~17 GB (4-bit ≈ 27 B × 0.57 GB + 1.5 GB ≈ 17 GB, estimate (fits a 24 GB GPU).)
API price
$0.42 in / $2.55 out per 1M tokens
Access
Free (open weights)
Last verified
18 Oct 2026

Price from openrouter.ai, 18 Oct 2026.

Benchmark results

BenchmarkResultEntry on the leaderboardAs ofSource
Artificial Analysis Intelligence Index34Qwen3.8 27B (Xhigh)
Raw value from the page's embedded model data; the site displays it rounded to an integer.
18 Oct 2026artificialanalysis.ai
Artificial Analysis MMMU-Pro (independent evaluation)76.3%Qwen3.8 27B (Xhigh)
Raw fraction 0.7630 from the model data embedded in https://artificialanalysis.ai/leaderboards/models, converted to percent; rank 46 of 124 entries with an MMMU-Pro result; no last-updated date shown.
19 Oct 2026artificialanalysis.ai
AutomationBench-AA (Artificial Analysis)48.2%Qwen3.8 27B (Xhigh)
AutomationBench-AA Score (share of each task's objectives completed, a task with any guardrail violation scoring zero); raw value from the page's embedded model data (displayed rounded); rank 28 of 32; Tasks Completed 15.68%, Objectives Completed 79.91%; no last-updated date shown, so as_of = retrieval date.
19 Oct 2026artificialanalysis.ai
EQ-Bench Creative Writing v31,671Qwen/Qwen3.8-27B
Elo score from the leaderboard data embedded in https://eqbench.com/creative_writing.js (file last modified 2026-09-24); rank 39 of 140; rubric score 15.50.
24 Sep 2026eqbench.com
LMArena Text Arena (overall, style control)1,438qwen3.8-27b
From LMArena's official leaderboard-dataset (text_style_control, split latest, category overall, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text (site blocks automated fetches).
15 Oct 2026huggingface.co
LMArena Text Arena: Creative Writing (style control)1,366qwen3.8-27b
From LMArena's official leaderboard-dataset (text_style_control, split latest, category creative_writing, published 2026-10-15), which mirrors https://arena.ai/leaderboard/text/creative-writing (site blocks automated fetches). Rank 160 of 414; 4668 votes; 95% CI 1356.8-1376.0.
15 Oct 2026huggingface.co
LMArena Vision Arena (overall, style control)1,244qwen3.8-27b
From LMArena's official leaderboard-dataset (vision_style_control, split latest, category overall, published 2026-10-09), which mirrors https://arena.ai/leaderboard/vision (site blocks automated fetches). Rank 61 of 165; 7013 votes; 95% CI 1236.0-1252.9.
9 Oct 2026huggingface.co
LMArena WebDev / Code Arena (overall)1,592Qwen3.8-27B
From LMArena's official leaderboard-dataset (webdev = Code Arena, split latest, category overall, published 2026-10-18), which mirrors https://arena.ai/leaderboard/webdev. Rank 30; 16172 votes; 95% CI 1585.5-1598.6.
18 Oct 2026huggingface.co
ParseBench (models)70.8%Qwen 3.8 27B (Thinking)
Overall (mean of tables, charts, content faithfulness, semantic formatting, visual grounding) from leaderboard.csv in the ParseBench repo (last leaderboard commit 2026-10-09). Non-thinking row scores 61.21.
9 Oct 2026github.com

Where you can use it

Earlier versions

  • Qwen3.5-27B · —

Sources