Tag
A user questions if others have noticed hallucinations and weird reasoning with the Qwen3.8-Flash-Next model on Mac, reporting issues with AppImage installation and tensor metadata from HuggingFace despite high quantization levels.
The article reports that the Qwen3.8-Flash-Next model achieves 120 tokens/second generation speed and 12k tokens/second prefill on a system with 4x AMD R9700 GPUs using optimized vLLM and a custom Docker image.
Support for the Qwen3.8-Flash-Next model has been merged into llama.cpp, enhancing its capabilities for local LLM inference in C/C++.