AI news ·
Daisytuner Raises €6 Million for Cross-Chip AI Optimization
Daisytuner raised €6 million and launched Daisyflow, an inference engine that compiles and tunes AI models across different hardware, claiming an 18% speed boost on AMD with Qwen 2.5.

Daisytuner has raised €6 million and released Daisyflow, an inference engine designed to compile and tune AI models across hardware from Nvidia, AMD, and other emerging chip vendors.
The company states that Daisyflow achieves an 18% speed improvement when running Qwen 2.5 on AMD hardware. Enterprises evaluating the engine should test this performance claim against their own workloads.