M5 Ultra 256GB or 2x DGX Spark? Which one?

Currently contemplating adding either M5 Ultra 256GB or dual DGX Spark in addition to existing 5090. I have been researching for past few days and have zeroed in on dual sparks because higher throughput for agentic coding / multiple parallel requests and prompt processing advantage over M5 ultra. Since the local LLM landscape changes everyday, it is difficult to find metrics generated by the same author. Ref: Mindstudio article comparing two . So far from what I have read, token generation is close match or slightly better with M5 ultra, whereas sparks take lead in prompt processing and multiple users/parallel requests. I am well aware that I will see 1/3rd of 5090 tokens speed with this choice but I will be able to run larger models. I am just not confident about the final choice for agentic coding - M5 Ultra 256GB or dual DGX Sparks? If you own either of these, can you please share the model you use, overall experience and basic metrics such as prompt processing and decode speeds? Thank you!