Qwen3.8-27B open-weight model tops Hugging Face trends

How this story unfolded

3 weeks · 3 reports · 137 community posts · 140 of 143 shown

1. Aug 11
   • Qwen 3.8-27b coming this week
2. Aug 13
   • 4 community posts
3. Aug 14
   • 26 community posts
4. Aug 15
   • 6 community posts
5. Aug 16
   • Qwen 3.8 27B is excellent, but it defaults to wildly overthinking thingssimonwillison.net
   • 10 community posts
6. Aug 17
   • 15 community posts
7. Aug 18
   • Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Indexsimonwillison.net
   • Qwen3.8-27B runs frontier-class coding agents and reasoning locally, no cloud API requiredventurebeat.com
   • 9 community posts
8. Aug 19
   • 7 community posts
9. Aug 20
   • 7 community posts
10. Aug 21
    • 5 community posts
11. Aug 22
    • 5 community posts
12. Aug 23
    • 7 community posts
13. Aug 24
    • 6 community posts
14. Aug 25
    • 3 community posts
15. Aug 26
    • 9 community posts
16. Aug 27
    • 2 community posts
17. Aug 28
    • 5 community posts
18. Aug 29
    • 3 community posts
19. Aug 30
    • 3 community posts
20. Aug 31
    • How I got Qwen 3.8 27b running at ~75t/s decode on 16GB RTX 5080
21. Sep 1
    • 2 community posts
22. Sep 2
    • How I got 280 tok/s on Qwen3.8 27B on 2xr9700's and 940k tokens kv cache