OpenWeightsTerminal

GLM 5.3 Flash

GLM 5.3 Flash

DGX Spark · EXL3 · EXL3

Decode · tok/s
18.6
Prefill
—
tok/100W
7.7
Full decode CUDA graph + MTP EAGLE k=2. 18.57 tok/s verified.

One box, not the 2× Mia recipe. Repo claim.

Officialclaimed
Open pair