20+ models
Qwen3.8 delivers substantial gains across coding, professional work, research, and long-horizon agentic tasks.
This experimental preview of the architecture that will underpin Qwen4.
pplx-decider is a 27B decision model from Perplexity, fine-tuned from Qwen3.8-27B
Qwen 3.8 27B, Distilled against Opus 5.5, GPT 6 Astra, and Grok 4.7
Qwen 3.8 Turbo abliterated w/ Heretic and distilled against Claude Fable
Qwen 3.8 distilled into Qwen 3.6 35B-A3B
Q2_K_S
RANA: an abliterated, uncensored Qwen3.8-27B with vision, thinking and tools, from Abliterant.
Text + MTP IQ3_S qwen3.8-flash-next-gsq-rco-gguf-mtp
pulled as BF16, vocab slice (combined_en_huge) requant to IQ4_NL (combined_en_huge) -- (English + Coding focus) - ~20% faster than Q4_K_M
This is a Qwen3.8 27B Model made for HYM3 Designs UI Version 5. Made for a 32GB RAM 8GB VRAM Devices or above. UI runs on Ubuntu 26.04, Python 3.14.7, Cuda 13.4.
Text + Vision + MTP Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-Q5_NVFP4
Verified configuration + measured results for abliterated (uncensored) Qwen3.8-27B running as an autonomous redteam/sysadmin agent, served via llama-swap. This card documents OUR tested configuration and results on live tasks
Reasoning model focused on shorter, token-efficient thinking + DFlash2 + map-k4v
Qwen3.8-27B-UD-Q4_K_M with fixed template for claude code