DeepSeek-V4-Flash hits 48.9 tok/s on single DGX Spark with speculative decoding

max_paperclips · x · 2026-08-03

Developer @danpacary successfully ran the 81GB DeepSeek-V4-Flash-0731 model on a single DGX Spark.

The test utilizes the ds4 inference engine developed by antirez.

Related event: DeepSeek-V4-Flash Local Deployment Tests: From Consumer GPUs to Pro Hardware(23 posts)→

Original post →

More from Infra

Infra channel →