Model Anatomy See what's inside open-weight AI models

Can I run Spark-X2.6-4B-FP8 on an RTX 5090?

Yes

Its published weights (4.45 GB) fit in the 28.8 GB an RTX 5090 can use, with room for ≈ 644K tokens of context.

Full analysis of Spark-X2.6-4B-FP8 Everything that runs on RTX 5090 Get the weights on Hugging Face
hf download XHToken/Spark-X2.6-4B-FP8

Files and context on RTX 5090

Each file of Spark-X2.6-4B-FP8 (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

FileSizeOn RTX 5090 Context (16-bit)Context (8-bit)
As published4.45 GBFits644K1M
4-bit (estimate)2.31 GBFits700K1M

Spark-X2.6-4B-FP8 on other hardware

Popular models that run on RTX 5090

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.