Model Release
+18.2%
Per-user speed vs FP8
GLM-5.3 · AMD · Inference · Quantization
GLM-5.3-Flash in MXFP4: A Lighter, Faster Checkpoint for AMD MI355X
We published an MXFP4 GLM-5.3-Flash checkpoint tuned for AMD Instinct MI355X. Against the official FP8 checkpoint on the same four GPUs, it is 40% smaller, with 12.4% more output throughput and an 18.2% faster P90 per-user rate (geometric means, c1–c48).
Read article