Muse Spark Open 816B — MI355X vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of MI355X (AMD CDNA 4) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on Muse Spark Open 816B. 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 →
MI355X posts 15001 tok/s/chip for $0.03 per million tokens at 197 tok/s/user on Muse Spark Open 816B; Vera Rubin NVL72 posts 83420 tok/s/chip for $0.01. Vera Rubin NVL72 is 131% cheaper per token; Vera Rubin NVL72 delivers 456% more tok/s/chip.
Throughput at 288 tok/s/user on Muse Spark Open 816B: MI355X hits 11787 tok/s/chip, Vera Rubin NVL72 hits 58970. Per-million costs land at $0.04 and $0.02 respectively. Vera Rubin NVL72 is 108% cheaper per token; Vera Rubin NVL72 delivers 400% more tok/s/chip.
MI355X / Vera Rubin NVL72 on Muse Spark Open 816B at 379 tok/s/user: 4410 / 46085 tok/s/chip, $0.09 / $0.02 per million tokens. Vera Rubin NVL72 is 334% cheaper per token; Vera Rubin NVL72 delivers 945% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI355X:15000.7Vera Rubin NVL72:83419.6 | MI355X:11787.0Vera Rubin NVL72:58969.7 | MI355X:4409.9Vera Rubin NVL72:46084.8 |
| Cost ($/M tok) | MI355X:$0.028Vera Rubin NVL72:$0.012 | MI355X:$0.035Vera Rubin NVL72:$0.017 | MI355X:$0.094Vera Rubin NVL72:$0.022 |
| tok/s/MW | MI355X:7177356Vera Rubin NVL72:25278662 | MI355X:5639724Vera Rubin NVL72:17869618 | MI355X:2110014Vera Rubin NVL72:13965102 |
| Concurrency | MI355X:~10Vera Rubin NVL72:~47 | MI355X:~12Vera Rubin NVL72:~30 | MI355X:~4Vera Rubin NVL72:~24 |