Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on GLM 5.3 744B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX →
Near the low end of the 87–157 tok/s/user interactivity band, at 104 tok/s/user on GLM 5.3 744B: GB300 NVL72 runs 25191 tok/s/chip at $0.03/M tokens, MI355X runs 5836 at $0.07/M. GB300 NVL72 is 180% cheaper per token; GB300 NVL72 delivers 332% more tok/s/chip.
Setting 122 tok/s/user as the target on GLM 5.3 744B, GB300 NVL72 produces 19027 tok/s/chip ($0.03 per million tokens) and MI355X produces 4528 ($0.09). GB300 NVL72 is 173% cheaper per token; GB300 NVL72 delivers 320% more tok/s/chip.
At 140 tok/s/user interactivity on GLM 5.3 744B, GB300 NVL72 delivers 14893 tok/s/chip at $0.04 per million tokens; MI355X delivers 1899 tok/s/chip at $0.22. GB300 NVL72 is 409% cheaper per token; GB300 NVL72 delivers 684% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:25191.3MI355X:5836.2 | GB300 NVL72:19027.0MI355X:4527.6 | GB300 NVL72:14893.5MI355X:1898.7 |
| Cost ($/M tok) | GB300 NVL72:$0.025MI355X:$0.071 | GB300 NVL72:$0.034MI355X:$0.092 | GB300 NVL72:$0.043MI355X:$0.219 |
| tok/s/MW | GB300 NVL72:11882707MI355X:2792446 | GB300 NVL72:8974998MI355X:2166304 | GB300 NVL72:7025225MI355X:908457 |
| Concurrency | GB300 NVL72:~254MI355X:~4 | GB300 NVL72:~84MI355X:~2 | GB300 NVL72:~54MI355X:~2 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
