OverviewAll LeaderboardsMethodologyComing soon
New Chat
Chat IndexQwen3.8-27B
Alibaba

Qwen3.8-27B

Qwen3.8-27B is an open source model from Alibaba. It ranks #54 of 65 models on the Chat Index with 993. Its best signal rank is #48 of 65, on Praise vs Complaint.

Chat Index Overview

Rank
#54

Rank spread 45 – 58.

Chat Index
993± 22
Pareto frontier

124 points under the frontier

Qwen3.8-27B
Developer
Alibaba
License
Apache 2.0
Token price (in / out)
$0.40 / $3 per 1M
Sessions
15,197
Ranking updated
Nov 13, 2026

Cost, Time, and Tokens

Cost per task$0.19
GPT-6 Luna (Low)$0.01AnthropicClaude Fable 5.2 (Max)$5.55

The median is $0.32.

Time per task4.1 min
MetaMuse Ember (Max)18 sAnthropicClaude Sonnet 5.5 (Max)15.2 min

The median is 3.4 min.

Output tokens per task26.3K
GPT-6 Luna (Low)2.1KAnthropicClaude Sonnet 5.5 (Max)135.6K

The median is 14.2K.

Chat Ranking

8009001,000
  1. 1,023
    Meta
    Muse Spark 1.1#47
  2. 1,020
    Qwen3.7-Plus#48
  3. 1,014
    Qwen3.8-Flash-Next#49
  4. 1,013
    MiniMax-M3.1-Flash#50
  5. 1,005
    Qwen3.7-Max#51
  6. 999
    Meta
    Muse Spark 1.2xHigh#52
  7. 999
    Stepfun
    Step 5 Preview#53
  8. 993
    Qwen3.8-27B#54
  9. 992
    Mistral Large 4#55
  10. 969
    Mistral Medium 3.5#56
  11. 961
    GPT-6 LunaLow#57
  12. 941
    Tencent
    Hy3#58
  13. 930
    MiniMax-M3#59
  14. 923
    Synlan
    Spark-X2.6#60
  15. 912
    Thinking Machines
    Inkling#61

Signal Performance

Qwen3.8-27BLeader
Rank on the board
  • Task CompletionHow often the model gets users to confirm the task is done.
    Claude Opus 5.5 (Max) · 1,370Claude Opus 5.5 (Max) · 1,370Qwen3.8-27B · 1,037
    #49of 65
  • Praise vs ComplaintHow often the model earns more explicitly positive responses than negative ones.
    Claude Fable 5.2 · 1,639Claude Fable 5.2 · 1,639Qwen3.8-27B · 980
    #48of 65
  • SteerabilityHow well the model lands user corrections when they push back.
    Claude Fable 5.2 · 1,246Claude Fable 5.2 · 1,246Qwen3.8-27B · 992
    #54of 65
  • Bash ReliabilityHow often the model’s bash commands succeed. Errors outside its control aren’t counted. E.g. network issues, permission denied, etc.
    Claude Opus 5.5 (Max) · 1,222Claude Opus 5.5 (Max) · 1,222Qwen3.8-27B · 932
    #61of 65
9001,0001,1001,2001,3001,4001,5001,6001,700

USE CASES

  • Chat with AI
  • Build Apps & Websites
  • Write & Edit Text
  • Search the Web
  • Generate Images
  • Generate Videos
  • Chose any model
  • Compare Models Side by Side

LEADERBOARD RANKINGS

  • Overall
  • Agent
  • Alignment Index
  • Text
  • WebDev
  • Image-to-WebDev
  • Text to Image
  • Image Edit
  • Text to Video
  • Image to Video
  • Video Edit
  • Vision
  • Document
  • Search

COMPANY

  • About Us
  • How It Works
  • Blog
  • Careers
  • Leaderboard Changelog
  • Product Changelog
  • Help Center
  • FAQ

LEGAL

  • Terms
  • Privacy
  • Cookies

FOLLOW

  • X
  • LinkedIn
  • YouTube
  • Discord

© Arena Intelligence