[untitled]1 points·by logotype·há 3 meses·0 comments1 commentsPost comment[–]logotype·há 3 mesesreplyTry it out and see how fast inference can be for agentic workflows. Works on CUDA and ROCm. Feedback appreciated!