Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and MI325X (AMD CDNA 3) on gpt-oss 120B. 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.
Throughput at 42 tok/s/user on gpt-oss 120B: MI300X hits 9176 tok/s/chip, MI325X hits 5626. Per-million costs land at $0.03 and $0.05 respectively. MI300X is 89% cheaper per token; MI300X delivers 63% more tok/s/chip.
MI300X / MI325X on gpt-oss 120B at 64 tok/s/user: 6666 / 3928 tok/s/chip, $0.04 / $0.08 per million tokens. MI300X is 97% cheaper per token; MI300X delivers 70% more tok/s/chip.
Toward the upper edge of the 21–107 tok/s/user interactivity band, at 85 tok/s/user on gpt-oss 120B: MI300X runs 5353 tok/s/chip at $0.05/M tokens, MI325X runs 2059 at $0.15/M. MI300X is 201% cheaper per token; MI300X delivers 160% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI300X:9176.3MI325X:5626.5 | MI300X:6666.1MI325X:3928.0 | MI300X:5352.6MI325X:2059.3 |
| Cost ($/M tok) | MI300X:$0.029MI325X:$0.054 | MI300X:$0.040MI325X:$0.078 | MI300X:$0.049MI325X:$0.148 |
| tok/s/MW | MI300X:6601669MI325X:3329264 | MI300X:4795752MI325X:2324277 | MI300X:3850782MI325X:1218509 |
| Concurrency | MI300X:~57MI325X:~16 | MI300X:~24MI325X:~8 | MI300X:~15MI325X:~12 |
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
gpt-oss 120B 8K / 1K Total Tokens per $1 TCO vs. Interactivity