Qwen3.8-27B Abliterated now available on Apple Silicon via MLX
SumitGup · x · 2026-08-16
Qwen3.8-27B Abliterated is now available for Apple Silicon using the MLX framework, offering 4-bit, 6-bit, 8-bit, and BF16 variants. The model has had refusal behaviors removed, scoring 0/100 on refusals while maintaining full quality and tool-use scores.
More from Infra
- Orbital Compute and Dyson Swarm GPU Clusters Predicted by 2030s — deanwball · 2026-08-16
- Help optimizing ComfyUI startup flags for low VRAM/AMD GPU — Morcas · 2026-08-16
- AFK Pilot Relay Open Sourced: Secure Message Routing for Coding Agents — PawelHuryn · 2026-08-16
- Faster Token Generation Drives Infinite Demand: Compute as the Ultimate Force Multiplier — OwariDa · 2026-08-16
- AMD ROCm vs NVIDIA CUDA: User Seeks Quality Comparison for Trellis2 Mesh Generation — Wake_Up_Morty · 2026-08-16
- Chinese firm tops Micron and Kioxia in NAND memory chip shipments — yogthos · 2026-08-16