BackendROCm
Decode98.47 tok/s
Prompt eval1510.64 tok/s
VRAM12 GB
Memory bandwidth~384 GB/s
ModelSpark-X2.5 4B
Parameters4.1B
QuantizationQ4_K_M
Context254K
KV quantf16
OffloadPartial (some layers in RAM)
About this result
Measured -p 512 -n 128 -r 3, per-test isolated telemetry (benchscope). Best decode (2.5x BF16), biggest context budget, coolest card (53 C); matched BF16 on every quality task. Full weights: 2.42 GiB.
Status: Measured in the VelsTech Lab on real hardware. Read the full Lab report →