Qwen 3.8 27B open-source model tops Hugging Face

How this story unfolded

4 weeks · 12 reports · 49 community posts · 61 of 68 shown

1. Aug 3
   • 2 community posts
2. Aug 11
   • Qwen 3.8-27b coming this week
3. Aug 12
   • Exact Qwen 3.8 27b release date and time
4. Aug 14
   • Qwen/Qwen3.8-27Bhuggingface.co
   • Qwen/Qwen3.8-27B-FP8huggingface.co
   • 8 community posts
5. Aug 15
   • orcarouter/Qwen3.8-27B-Uncensored-FP8huggingface.co
   • 4 community posts
6. Aug 16
   • Qwen 3.8 27B is excellent, but it defaults to wildly overthinking thingssimonwillison.net
   • 2 community posts
7. Aug 17
   • 0bserverx/Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUFhuggingface.co
   • AtomicChat/Qwen3.8-27B-GGUFhuggingface.co
   • DavidAU/Qwen3.8-27B-Cold-Fusion-GAIN-V1.1huggingface.co
   • 8 community posts
8. Aug 18
   • orcarouter/Qwen3.8-27B-Uncensored-MLXhuggingface.co
   • 4 community posts
9. Aug 19
   • Qwen3.8-27B on 2x 3090 + vLLM + DFlash2: 218 tok/s single request
10. Aug 20
    • 2 community posts
11. Aug 21
    • 2 community posts
12. Aug 23
    • outsourc-e/Qwen3.8-27B-Unleashed-GGUFhuggingface.co
    • 2 community posts
13. Aug 24
    • This Small AI Will Change Everythingyoutube.com
    • 3 community posts
14. Aug 26
    • 2 community posts
15. Aug 27
    • okay this is too cute 😭 i saw Microduck and immediately asked Qwen3.8-27B to build a tiny browser...
16. Aug 28
    • 2 community posts
17. Aug 29
    • DavidAU/Qwen3.8-27B-Cold-Fable-Fusion-GAIN-V1.1-732-Heretic-Uncensored-stage1huggingface.co
    • Qwen 3.8 27B at 50 tok/s with 100k Context on a 16GB GPU! (beellama.cpp)
18. Aug 30
    • 2 community posts
19. Aug 31
    • ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUFhuggingface.co
    • Qwen 3.8 27B - Fantastic German capabilities