Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on gpt-oss 120B. 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.
Near the low end of the 68β305 tok/s/user interactivity band, at 127 tok/s/user on gpt-oss 120B: GB200 NVL72 runs 39042 tok/s/chip at $0.01/M tokens, MI355X runs 19134 at $0.02/M. GB200 NVL72 is 65% cheaper per token; GB200 NVL72 delivers 104% more tok/s/chip.
Setting 187 tok/s/user as the target on gpt-oss 120B, GB200 NVL72 produces 28526 tok/s/chip ($0.02 per million tokens) and MI355X produces 11474 ($0.04). GB200 NVL72 is 100% cheaper per token; GB200 NVL72 delivers 149% more tok/s/chip.
At 246 tok/s/user interactivity on gpt-oss 120B, GB200 NVL72 delivers 18958 tok/s/chip at $0.03 per million tokens; MI355X delivers 3618 tok/s/chip at $0.12. GB200 NVL72 is 323% cheaper per token; GB200 NVL72 delivers 424% more tok/s/chip at this point. (Numbers reflect the default 8k/1k Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:39042.3MI355X:19133.8 | GB200 NVL72:28525.7MI355X:11474.5 | GB200 NVL72:18958.0MI355X:3618.0 |
| Cost ($/M tok) | GB200 NVL72:$0.013MI355X:$0.022 | GB200 NVL72:$0.018MI355X:$0.036 | GB200 NVL72:$0.027MI355X:$0.115 |
| tok/s/MW | GB200 NVL72:20878221MI355X:9154931 | GB200 NVL72:15254396MI355X:5490178 | GB200 NVL72:10137992MI355X:1731095 |
| Concurrency | GB200 NVL72:~489MI355X:~18 | GB200 NVL72:~109MI355X:~7 | GB200 NVL72:~53MI355X:~16 |
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 β’ Drag to pan β’ Double-click to reset β’ 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.
gpt-oss 120B 8K / 1K Total Tokens per $1 TCO vs. Interactivity