MiniMax M2.5/M2.7 — B300 vs MI325X
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) on MiniMax M2.5/M2.7. 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.
At 35 tok/s/user interactivity on MiniMax M2.5/M2.7, B300 delivers 11471 tok/s/chip at $0.05 per million tokens; MI325X delivers 2647 tok/s/chip at $0.12. B300 is 111% cheaper per token; B300 delivers 333% more tok/s/chip at this point.
B300 posts 7774 tok/s/chip for $0.08 per million tokens at 56 tok/s/user on MiniMax M2.5/M2.7; MI325X posts 1814 tok/s/chip for $0.17. B300 is 109% cheaper per token; B300 delivers 329% more tok/s/chip.
Throughput at 78 tok/s/user on MiniMax M2.5/M2.7: B300 hits 4363 tok/s/chip, MI325X hits 1009. Per-million costs land at $0.14 and $0.30 respectively. B300 is 110% cheaper per token; B300 delivers 332% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:11471.3MI325X:2647.3 | B300:7774.5MI325X:1814.3 | B300:4362.6MI325X:1009.1 |
| Cost ($/M tok) | B300:$0.055MI325X:$0.115 | B300:$0.081MI325X:$0.168 | B300:$0.144MI325X:$0.303 |
| tok/s/MW | B300:6037520MI325X:1566421 | B300:4091823MI325X:1073566 | B300:2296114MI325X:597118 |
| Concurrency | B300:~521MI325X:~18 | B300:~335MI325X:~8 | B300:~6MI325X:~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
MiniMax M2.5/2.7 230B • 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