hugpadhugpad

DavidAU / Qwen3.8-Flash-Next-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF

Vision99.9K38Updated Oct 21, 2026
Hugging Face

Downloads

99.9K

Likes

38

Last updated

Oct 21, 2026

License

apache-2.0

- 180 Billion parameters with 512 experts, 10 activated per token. - Turbo Brilliance: Token reduction (thinking) when using AI CONSTRUCTS (preloads at 1000+ t/s) : EXTREME ; instruct modes: 0 thinking tokens. - SOTA BF16 performance in "IQ3S" using quant tuning stronger than …

Tasks & library

  • image text to text

Tags

  • gguf
  • 27 reasoning modes
  • 27 instruct modes
  • 15 AI Constructs
  • 5 Super AI Constructs
  • Brainstorm AI Construct

Technical specifications

ParametersNot available
LibraryNot available
Languages16 languages
Base modelISTA-DASLab/Qwen3.8-Flash-Next-GSQ-RCO-GGUF
Licenseapache-2.0