Intel’s LLM-Scaler project that makes it easy to deploy various large language models on modern Arc Graphics hardware is out with a new test release to expand its LLM coverage.
Intel on Thursday released llm-scaler-vllm 0.14.0-b8.1 as the latest version of this Docker-based deployment setup for LLMs on Intel graphics hardware leveraging the excellent vLLM. Ultimately this is building off and benefiting from Intel’s work over the past year with Project Battlematrix driver enhancements.
With this new LLM-Scaler-vLLM release there is support now for more Qwen models on Intel hardware. New support includes Qwen3.5-27B, Qwen3.5-35B-A3B and Qwen3.5-122B-A10B (FP8 and INT4). Qwen3-ASR-1.7B is also now supported by this Intel open-source software stack too.
Downloads and more details on this llm-scaler-vllm release via GitHub.
