Trending storyDeveloping

Artificial Analysis ranks AMD MI355X between H200 and B200 on GLM-5.2/5.3 serving

2 articles1 sourcesince Oct 14Last article 53m ago · Oct 14, 1:12 PM ET

Overview

AISummary of 2 articles

Artificial Analysis says AMD's MI355X serves about 26M tokens/s per MW at 100 tok/s per user on GLM-5.2/5.3.

That is ahead of NVIDIA's H200 at about 16M but behind the B200 at about 42M and the B300 at about 44M.

For single-agent speed, the MI355X reaches 228 tok/s, matching the H200. Artificial Analysis reports 312 tok/s on the B200 and 333 tok/s on the B300.

Written by AI from the articles below · updated Oct 14, 1:21 PM ET

Check the sources:

Latest developmentOct 14, 1:12 PM ET

The articles in this story. Times are ET.

Oct 14

1:12 PM
Artificial Analysis@ArtificialAnlysOfficial
AMD MI355X lands between Hopper and Blackwell on GLM serving
AIArtificial Analysis says AMD's MI355X serves about 26M tokens/s per MW at 100 tok/s per user on GLM-5.2/5.3. That beats H200 at about 16M but trails B200 at about 42M and B300 at about 44M. Single-agent speed is 228 tok/s, matching H200 and below B200 at 312 and B300 at 333.
1:03 PM
Artificial Analysis@ArtificialAnlysOfficial
AMD MI355X lands between Hopper and Blackwell on GLM serving
AIArtificial Analysis says AMD's MI355X serves about 26M tokens/s per MW at 100 tok/s per user on GLM-5.2/5.3. That beats H200 at about 16M but trails B200 at about 42M and B300 at about 44M. Single-agent speed is 228 tok/s, matching H200 and below B200 at 312 and B300 at 333.

Not enough continuous observations to show a trend yet.