Head-to-head AI inference benchmark comparison of MI355X (AMD CDNA 4) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeekv4 Pro 0813 1.6T. 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 →
Setting 110 tok/s/user as the target on DeepSeekv4 Pro 0813 1.6T, MI355X produces 19563 tok/s/chip ($0.02 per million tokens) and Vera Rubin NVL72 produces 273427 ($0.00). Vera Rubin NVL72 is 481% cheaper per token; Vera Rubin NVL72 delivers 1298% more tok/s/chip.
At 145 tok/s/user interactivity on DeepSeekv4 Pro 0813 1.6T, MI355X delivers 10316 tok/s/chip at $0.04 per million tokens; Vera Rubin NVL72 delivers 235101 tok/s/chip at $0.00. Vera Rubin NVL72 is 847% cheaper per token; Vera Rubin NVL72 delivers 2179% more tok/s/chip at this point.
MI355X posts 4299 tok/s/chip for $0.10 per million tokens at 181 tok/s/user on DeepSeekv4 Pro 0813 1.6T; Vera Rubin NVL72 posts 153620 tok/s/chip for $0.01. Vera Rubin NVL72 is 1385% cheaper per token; Vera Rubin NVL72 delivers 3474% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI355X:19563.5Vera Rubin NVL72:273426.8 | MI355X:10315.7Vera Rubin NVL72:235100.8 | MI355X:4298.8Vera Rubin NVL72:153619.6 |
| Cost ($/M tok) | MI355X:$0.021Vera Rubin NVL72:$0.004 | MI355X:$0.040Vera Rubin NVL72:$0.004 | MI355X:$0.097Vera Rubin NVL72:$0.007 |
| tok/s/MW | MI355X:9360509Vera Rubin NVL72:82856601 | MI355X:4935760Vera Rubin NVL72:71242673 | MI355X:2056836Vera Rubin NVL72:46551396 |
| Concurrency | MI355X:~39Vera Rubin NVL72:~2904 | MI355X:~20Vera Rubin NVL72:~2516 | MI355X:~6Vera Rubin NVL72:~1458 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
1 The ATOM engine is promising, however it has yet to serve production tokens. It is still in its infant stage.
DeepSeek V4 Pro 0813 1.6T 8K / 1K Total Tokens per $1 TCO vs. Interactivity