MI355X (ROCm/SGLang FP8) cost-per-Mtoken advantage vs B200 (SGLang) at points on the GLM-5 single-node frontier — gap reversed on funded workloads
~40%observed
| Value kind | observed — Reported measurements, counts, and specifications keep the precision and scope stated by their source; an exact specification is not treated as a range. |
|---|---|
| Scope | MI355X cost/Mtoken gap |
| As of | 2026 |
| Source | SemiAnalysis InferenceX · GLM-5 FP8, ISL 8192/OSL 1024, single node: peak 1.41x at 18 tok/s/user w/ MTP ($0.30/M B200 vs $0.22/M MI355X, ~40%); advantage holds ~10-77 tok/s/user, B200 back ahead >90 |
| Review | checking…review by 2026-09-27 · fast cadence |
| Recorded changes | last 2026-06-29 |
| Claim id | mi355x-rocm-sglang-fp8-cost-per-mtoken |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.