Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Kimi K3 2.8T. 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 →
GB300 NVL72 / MI355X on Kimi K3 2.8T at 36 tok/s/user: 12503 / 7034 tok/s/chip, $0.05 / $0.06 per million tokens. GB300 NVL72 is 15% cheaper per token; GB300 NVL72 delivers 78% more tok/s/chip.
Around the middle of the 9–120 tok/s/user interactivity band, at 64 tok/s/user on Kimi K3 2.8T: GB300 NVL72 runs 11795 tok/s/chip at $0.05/M tokens, MI355X runs 3443 at $0.12/M. GB300 NVL72 is 122% cheaper per token; GB300 NVL72 delivers 243% more tok/s/chip.
Setting 92 tok/s/user as the target on Kimi K3 2.8T, GB300 NVL72 produces 9074 tok/s/chip ($0.07 per million tokens) and MI355X produces 1860 ($0.22). GB300 NVL72 is 217% cheaper per token; GB300 NVL72 delivers 388% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:12503.4MI355X:7033.7 | GB300 NVL72:11794.8MI355X:3443.0 | GB300 NVL72:9074.3MI355X:1860.1 |
| Cost ($/M tok) | GB300 NVL72:$0.051MI355X:$0.059 | GB300 NVL72:$0.054MI355X:$0.121 | GB300 NVL72:$0.071MI355X:$0.224 |
| tok/s/MW | GB300 NVL72:5897823MI355X:3365424 | GB300 NVL72:5563608MI355X:1647362 | GB300 NVL72:4280332MI355X:890001 |
| Concurrency | GB300 NVL72:~55MI355X:~21 | GB300 NVL72:~48MI355X:~5 | GB300 NVL72:~46MI355X:~4 |
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.
