Llama 3.3 70B — B200 vs MI325X
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) on Llama 3.3 70B. 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.
Setting 37 tok/s/user as the target on Llama 3.3 70B, B200 produces 6594 tok/s/chip ($0.07 per million tokens) and MI325X produces 1712 ($0.18). B200 is 145% cheaper per token; B200 delivers 285% more tok/s/chip.
At 61 tok/s/user interactivity on Llama 3.3 70B, B200 delivers 4854 tok/s/chip at $0.10 per million tokens; MI325X delivers 1301 tok/s/chip at $0.23. B200 is 137% cheaper per token; B200 delivers 273% more tok/s/chip at this point.
B200 posts 3586 tok/s/chip for $0.13 per million tokens at 84 tok/s/user on Llama 3.3 70B; MI325X posts 734 tok/s/chip for $0.42. B200 is 211% cheaper per token; B200 delivers 389% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:6593.5MI325X:1711.8 | B200:4854.3MI325X:1300.5 | B200:3586.3MI325X:733.9 |
| Cost ($/M tok) | B200:$0.073MI325X:$0.179 | B200:$0.099MI325X:$0.235 | B200:$0.134MI325X:$0.416 |
| tok/s/MW | B200:3855848MI325X:1012893 | B200:2838777MI325X:769539 | B200:2097233MI325X:434251 |
| Concurrency | B200:~44MI325X:~24 | B200:~32MI325X:~22 | B200:~22MI325X:~8 |
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
Llama 3.3 70B Instruct • 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