Hot French startup ZML releases free product to speed inference across lots of AI chips

French startup ZML, known for its AI infrastructure work and backed by Turing Award winner Yann LeCun, has released a free software tool called ZML/LLMD. The tool is designed to accelerate inference across diverse AI chips, potentially lowering operational costs for developers running large language models and other AI workloads. According to TechCrunch AI, the software optimizes model execution on multiple hardware types, from GPUs to custom accelerators, without requiring significant code changes from users. This release comes as the AI industry seeks more efficient ways to deploy models, given the high cost of inference at scale. For developers and solopreneurs building AI workflows, ZML/LLMD offers a way to reduce infrastructure expenses and improve performance without vendor lock-in. The company has made the tool freely available, aiming to attract a broad user base and contribute to the open-source ecosystem. Practical implications include faster response times for user-facing applications and the ability to run models on more affordable hardware. However, adoption may depend on ease of integration and community support.