French startup ZML launches free multi-chip AI inference tool

ZML, a Paris-based AI startup backed by Yann LeCun and $20 million in venture funding, has launched LLMD, a free LLM inference server designed to run open-source AI models across multiple chip types including Nvidia, AMD, Google TPU, Apple Metal, and Intel Arc. The tool aims to break vendor lock-in and maximize performance across diverse hardware. The release comes amid a booming inference market, where ZML competes with players like Baseten and Inferact. Unlike its open-source predecessor, LLMD launches as a free product while ZML gauges usage before monetizing. The 20-person team also claims to be co-designing silicon with chip partners, many of them European.