Head-to-head AI inference benchmark comparison of MI355X (AMD CDNA 4) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on Kimi K3 2.8T. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β 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 β
MI355X / Vera Rubin NVL72 on Kimi K3 2.8T at 121 tok/s/user: 3594 / 40998 tok/s/chip, $0.12 / $0.02 per million tokens. Vera Rubin NVL72 is 374% cheaper per token; Vera Rubin NVL72 delivers 1041% more tok/s/chip.
Around the middle of the 103β177 tok/s/user interactivity band, at 140 tok/s/user on Kimi K3 2.8T: MI355X runs 1745 tok/s/chip at $0.24/M tokens, Vera Rubin NVL72 runs 37279 at $0.03/M. Vera Rubin NVL72 is 788% cheaper per token; Vera Rubin NVL72 delivers 2037% more tok/s/chip.
Setting 159 tok/s/user as the target on Kimi K3 2.8T, MI355X produces 1691 tok/s/chip ($0.25 per million tokens) and Vera Rubin NVL72 produces 33784 ($0.03). Vera Rubin NVL72 is 730% cheaper per token; Vera Rubin NVL72 delivers 1898% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI355X:3593.6Vera Rubin NVL72:40997.5 | MI355X:1744.8Vera Rubin NVL72:37279.0 | MI355X:1691.1Vera Rubin NVL72:33784.1 |
| Cost ($/M tok) | MI355X:$0.116Vera Rubin NVL72:$0.024 | MI355X:$0.239Vera Rubin NVL72:$0.027 | MI355X:$0.246Vera Rubin NVL72:$0.030 |
| tok/s/MW | MI355X:1719411Vera Rubin NVL72:12423489 | MI355X:834846Vera Rubin NVL72:11296664 | MI355X:809123Vera Rubin NVL72:10237593 |
| Concurrency | MI355X:~9Vera Rubin NVL72:~229 | MI355X:~11Vera Rubin NVL72:~202 | MI355X:~7Vera Rubin NVL72:~176 |
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.
