DeepSeekv4 Pro 0813 1.6T — GB300 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) 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 →
GB300 NVL72 / MI355X on DeepSeekv4 Pro 0813 1.6T at 55 tok/s/user: 107320 / 8477 tok/s/chip, $0.01 / $0.05 per million tokens. GB300 NVL72 is 722% cheaper per token; GB300 NVL72 delivers 1166% more tok/s/chip.
Around the middle of the 34–117 tok/s/user interactivity band, at 76 tok/s/user on DeepSeekv4 Pro 0813 1.6T: GB300 NVL72 runs 69128 tok/s/chip at $0.01/M tokens, MI355X runs 4880 at $0.09/M. GB300 NVL72 is 820% cheaper per token; GB300 NVL72 delivers 1317% more tok/s/chip.
Setting 97 tok/s/user as the target on DeepSeekv4 Pro 0813 1.6T, GB300 NVL72 produces 40182 tok/s/chip ($0.02 per million tokens) and MI355X produces 3368 ($0.12). GB300 NVL72 is 675% cheaper per token; GB300 NVL72 delivers 1093% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:107320.0MI355X:8477.4 | GB300 NVL72:69128.1MI355X:4879.7 | GB300 NVL72:40182.1MI355X:3368.4 |
| Cost ($/M tok) | GB300 NVL72:$0.006MI355X:$0.049 | GB300 NVL72:$0.009MI355X:$0.085 | GB300 NVL72:$0.016MI355X:$0.124 |
| tok/s/MW | GB300 NVL72:50622659MI355X:4056192 | GB300 NVL72:32607599MI355X:2334778 | GB300 NVL72:18953815MI355X:1611688 |
| Concurrency | GB300 NVL72:~1106MI355X:~32 | GB300 NVL72:~859MI355X:~8 | GB300 NVL72:~381MI355X:~5 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
Total Tokens per $1 USD (Owning - Hyperscaler) vs. Interactivity
DeepSeek V4 Pro 1.6T • FP4 • 8K / 1K • Source: SemiAnalysis InferenceX™
TCO $/chip/hr: H100: 1.17H200: 1.22B200: 1.73B300: 2.26GB200: 1.86GB300: 2.31MI300X: 0.95MI325X: 1.1MI355X: 1.5RTX6000PRO: 0.68
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.
No data available
Please change the model, sequence, precision, date range or chip selection.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip