Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI300X (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 β same interactions as the main inference chart.
Near the low end of the 14β101 tok/s/user interactivity band, at 36 tok/s/user on Llama 3.3 70B: B200 runs 6670 tok/s/chip at $0.07/M tokens, MI300X runs 1823 at $0.14/M. B200 is 101% cheaper per token; B200 delivers 266% more tok/s/chip.
Setting 58 tok/s/user as the target on Llama 3.3 70B, B200 produces 5052 tok/s/chip ($0.10 per million tokens) and MI300X produces 1144 ($0.23). B200 is 142% cheaper per token; B200 delivers 341% more tok/s/chip.
At 80 tok/s/user interactivity on Llama 3.3 70B, B200 delivers 3781 tok/s/chip at $0.13 per million tokens; MI300X delivers 679 tok/s/chip at $0.39. B200 is 206% cheaper per token; B200 delivers 457% more tok/s/chip at this point. (Numbers reflect the default 8k/1k Β· fp8 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:6669.9MI300X:1822.9 | B200:5051.6MI300X:1144.4 | B200:3780.8MI300X:679.1 |
| Cost ($/M tok) | B200:$0.072MI300X:$0.145 | B200:$0.095MI300X:$0.231 | B200:$0.127MI300X:$0.389 |
| tok/s/MW | B200:3900512MI300X:1311463 | B200:2954131MI300X:823315 | B200:2211020MI300X:488526 |
| Concurrency | B200:~47MI300X:~27 | B200:~32MI300X:~16 | B200:~25MI300X:~8 |
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
Llama 3.3 70B Instruct 8K / 1K Total Tokens per $1 TCO vs. Interactivity