Qwen 3.8
Best conversion per model
One row per base model. The full list of attempts is on each model page.
| # | Model | Author | Params | KL | Refusals | Score | Evidence | Downloads |
|---|---|---|---|---|---|---|---|---|
| 1 | DeepSeek-R1-0528-Qwen3-8B deepseek-ai | saidutta69 | 8B | 0.0082 | 3/1003% | 98.1 | BestParetoGGUFSAFETENSORS | 1.5K |
| 2 | Qwen3.8-27B Qwen | trohrbaugh | 27B | 0.053 | 0/1000% | 97.5 | BestEXLFP8GGUF | 282.8K |
| 3 | Qwen3.8-27B-Ultra-Uncensored-Heretic-Native-MTP-Preserved llmfan46 | llmfan46 | 27B | 0.024 | 3/1003% | 97.3 | BestParetoGGUFGPTQNVFP4 | 115.7K |
| 4 | Qwen3.8-Flash-Next Qwen | trohrbaugh | 180B | 0.082 | 0/1000% | 96.2 | BestParetoGGUFMLXSAFETENSORS | 28K |
| 5 | Qwen3.8-27B-FP8 Qwen | krthkyadtya | 27B | 0.083 | 0/300% | 96.2 | BestParetoFP8SAFETENSORS | 146 |
| 6 | Qwen3.8-9B-Distill empero-ai | petruhonk | 9B | 0.031 | 6/1006% | 95.5 | BestParetoGGUFSAFETENSORS | 9K |
| 7 | ThinkingCap-Qwen3.8-27B bottlecapai | IstroSec | 27B | 0.065 | 6/1006% | 93.9 | BestParetoFP8GGUFSAFETENSORS | 235 |
| 8 | Qwen3-8B Qwen | jbeslt | 8B | 0.060 | 7/1007% | 93.7 | BestParetoSAFETENSORS | 193 |
| 9 | Qwen3.8-27B-Fable-Distill TeichAI | armand0e | 27B | 0.100 | 5/1005% | 92.9 | BestParetoGGUFSAFETENSORS | 23.7K |
| 10 | Qwen3.8-Flash-Next-exl3 turboderp | jeffpeng3 | 42.5B | 0.056 | 54/10054% | 70.3 | BestParetoEXLSAFETENSORS | 172 |
| 11 | Qwen3.8-9B empero-ai | MegaPanchamZ | 9B | 0.014 | 99/10099% | 49.8 | BestParetoGGUFLORASAFETENSORS | 11.7K |
All models
Unmeasured models
Huihui-Qwen3.8-27B-abliterated, Qwen3-8B-HiveMind-Heretic-qx86-hi-mlx, Qwen3.5-4B-EmperoAI-Qwen3.8-Distill-Heretic-Abliterated-MTP, Qwen3.8-27B, Qwen3.8-27B-ABLITERATED-UNCENSORED-PHILADELPHIA-CLASS, Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-BF16, qwen3.8-27b-aragon-uncensored, Qwen3.8-27B-Cold-Fable-Fusion-GAIN-V1.1-732-Heretic-Uncensored-stage1, Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-heretic, Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-heretic-NOESIS-BF16, Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-heretic-W4A16, Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2-GGUF, Qwen3.8-27B-EXL3-3.5bpw, Qwen3.8-27B-heretic, Qwen3.8-27B-heretic, Qwen3.8-27B-Max-Ablit-Heretic, Qwen3.8-27B-Optimal-Ablit-Heretic, Qwen3.8-27B-OrcaRouter-Uncensored-llamafile, Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX, Qwen3.8-27B-Ultra-Uncensored-Heretic-Native-MTP-Preserved-mlx-8Bit, Qwen3.8-27B-Uncensored-GGUF, Qwen3.8-27B-Uncensored-Heretic-Abliterated, Qwen3.8-2B, Qwen3.8-2B-Distill, Qwen3.8-35B-A3B-Distill-oQ8-fp16-mtp, Qwen3.8-4B, Qwen3.8-4B-Distill, Qwen3.8-4B-Distill-GGUF, Qwen3.8-9B-heretic-uncensored, Qwen3.8-Flash-Next-Uncensored, Swift-1.5-Qwen3.8-27b, Swift-1.5-Qwen3.8-27B-Uncensored-MTP-VAL-APEX-I-MiniPlus-V3-GGUF — conversions exist but their publishers did not report KL and refusal numbers, so they are not ranked.