Qwen3.8-27B-antislop · GGUF

Try it (Space) · Main model · rawmodels.ai · Original paper: unslop / RL-XAR (Meta FAIR) · RL environment

GGUF builds of Qwen3.8-27B-antislop: Qwen3.8-27B trained to write less like an LLM. #4 of 29 on EQ-Bench 4 (Elo 1320), above GPT-5.5 and Claude Opus 4.7/4.8.

Text only, llama.cpp efe3626, static quants. Hashes in weights_manifest.json.

llama-cli -m Qwen3.8-27B-antislop-Q4_K_M.gguf -p "Finish this tweet naturally: Monday mornings are" -n 64 --temp 0.8

Method, benchmarks and how to reproduce are on the main model card. The reward is public as an RL environment: rawmodels/unslop-harbor.