Re: undefined #2 Post by logotype » Mon, Apr 13, 2026, 8:54 PM UTC Try it out and see how fast inference can be for agentic workflows. Works on CUDA and ROCm. Feedback appreciated!