SGLang Supports Inkling-Small on Dual DGX Spark, Hits 24 tok/s

ying11231 · x · 2026-08-01

SGLang has officially announced support for running Inkling-Small across 2× NVIDIA DGX Spark systems connected via ConnectX-7.

This setup enables strong agentic capabilities on a compact local hardware platform. The current configuration achieves a generation speed of 24 tok/s (concurrency=1) without utilizing MTP.

Original post →

More from coding & agent

coding & agent channel →