Intel LLM-Scaler Ready With Muse Glimmer Support, Other LLMs & Features - Phoronix

Intel has released updates to its LLM-Scaler project, designed to facilitate running generative AI on Arc B-Series graphics cards using tools like vLLM and ComfyUI. The latest LLM-Scaler-vLLM beta introduces same-day support for Meta's Muse Glimmer 30B model with FP8 online quantization on hardware like the Arc Pro B70, alongside DFlash support and performance improvements for Gemma and Qwen models. Additionally, the LLM-Scaler-Omni beta updates the ComfyUI XPU stack, adds support for local video generation tools like MiniMax H3 and Wan Animate 2, and broadens optimized model coverage.

(Source：Phoronix)