Head-to-head AI inference benchmark comparison of MI325X (AMD CDNA 3) and MI355X (AMD CDNA 4) 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.
Near the low end of the 14–107 tok/s/user interactivity band, at 37 tok/s/user on Llama 3.3 70B: MI325X runs 1712 tok/s/chip at $0.18/M tokens, MI355X runs 4326 at $0.10/M. MI355X is 85% cheaper per token; MI355X delivers 153% more tok/s/chip.
Setting 61 tok/s/user as the target on Llama 3.3 70B, MI325X produces 1301 tok/s/chip ($0.23 per million tokens) and MI355X produces 2613 ($0.16). MI355X is 47% cheaper per token; MI355X delivers 101% more tok/s/chip.
At 84 tok/s/user interactivity on Llama 3.3 70B, MI325X delivers 734 tok/s/chip at $0.42 per million tokens; MI355X delivers 1086 tok/s/chip at $0.38. MI355X is 8% cheaper per token; MI355X delivers 48% more tok/s/chip at this point. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI325X:1711.8MI355X:4326.2 | MI325X:1300.5MI355X:2612.5 | MI325X:733.9MI355X:1085.7 |
| Cost ($/M tok) | MI325X:$0.179MI355X:$0.096 | MI325X:$0.235MI355X:$0.159 | MI325X:$0.416MI355X:$0.384 |
| tok/s/MW | MI325X:1012893MI355X:2069965 | MI325X:769539MI355X:1250014 | MI325X:434251MI355X:519473 |
| Concurrency | MI325X:~24MI355X:~18 | MI325X:~22MI355X:~22 | MI325X:~8MI355X:~6 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr: VR200: 3.61H100: 1.17H200: 1.22B200: 1.73B300: 2.26GB200: 1.86GB300: 2.31MI300X: 0.95MI325X: 1.1MI355X: 1.5RTX6000PRO: 0.68Jalapeño (Teacup): 1.47
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
Llama 3.3 70B Instruct 8K / 1K Total Tokens per $1 TCO vs. Interactivity