Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on GLM 5/5.1. 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.
Throughput at 36 tok/s/user on GLM 5/5.1: GB200 NVL72 hits 10443 tok/s/chip, MI355X hits 1218. Per-million costs land at $0.05 and $0.34 respectively. GB200 NVL72 is 591% cheaper per token; GB200 NVL72 delivers 757% more tok/s/chip.
GB200 NVL72 / MI355X on GLM 5/5.1 at 43 tok/s/user: 10265 / 949 tok/s/chip, $0.05 / $0.44 per million tokens. GB200 NVL72 is 773% cheaper per token; GB200 NVL72 delivers 982% more tok/s/chip.
Toward the upper edge of the 30–56 tok/s/user interactivity band, at 50 tok/s/user on GLM 5/5.1: GB200 NVL72 runs 9984 tok/s/chip at $0.05/M tokens, MI355X runs 670 at $0.62/M. GB200 NVL72 is 1101% cheaper per token; GB200 NVL72 delivers 1390% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:10442.9MI355X:1218.0 | GB200 NVL72:10265.2MI355X:948.6 | GB200 NVL72:9983.9MI355X:670.1 |
| Cost ($/M tok) | GB200 NVL72:$0.049MI355X:$0.342 | GB200 NVL72:$0.050MI355X:$0.439 | GB200 NVL72:$0.052MI355X:$0.622 |
| tok/s/MW | GB200 NVL72:5584446MI355X:582775 | GB200 NVL72:5489410MI355X:453886 | GB200 NVL72:5338994MI355X:320643 |
| Concurrency | GB200 NVL72:~2253MI355X:~8 | GB200 NVL72:~1843MI355X:~8 | GB200 NVL72:~1340MI355X:~7 |
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.
GLM5/5.1 744B 8K / 1K Total Tokens per $1 TCO vs. Interactivity