-
qwen38-27b-turbofcfusion
DavidAU's uncensored multi-stage decensor of Qwen Team's Qwen3.8-27B: built-in MTP, native vision, 262K ctx (1M ext), thinking by default. Q4_K_M GGUF. Dual TITAN RTX @96.8K: 40 tok/s decode, acceptance 0.70. Cred: Qwen Team + DavidAU.
tools thinking701 Pulls 1 Tag Updated 2 weeks ago
-
wen36-35b-uncensored-1m
Qwen3.6-35B-A3B Uncensored is a Mixture-of-Experts model with 35.5B total parameters and 3B activated, extended to 1M-token context with the official MTP speculative-decoding layer, delivering 1.5x decode speedup on Ollama, with native thinking and tools.
tools thinking678 Pulls 2 Tags Updated 1 month ago