MiniMax M3.1 Flash 428B · Chip comparison
MiniMax M3.1 Flash 428B — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on MiniMax M3.1 Flash 428B. 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.
AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX →
At 200 tok/s/user interactivity on MiniMax M3.1 Flash 428B, GB200 NVL72 delivers 98,848 tok/s/chip at $0.005 per million tokens; MI355X delivers 80,133 tok/s/chip at $0.005. Cost per token is essentially tied; GB200 NVL72 delivers 23% more tok/s/chip at this point.
GB200 NVL72 posts 24,246 tok/s/chip for $0.021 per million tokens at 340 tok/s/user on MiniMax M3.1 Flash 428B; MI355X posts 36,632 tok/s/chip for $0.011. MI355X is 47% cheaper per token; MI355X delivers 51% more tok/s/chip.
Throughput at 481 tok/s/user on MiniMax M3.1 Flash 428B: GB200 NVL72 hits 10,580 tok/s/chip, MI355X hits 17,494. Costs per million tokens land at $0.049 and $0.024 respectively. MI355X is 51% cheaper per token; MI355X delivers 65% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL7298,848MI355X80,133 | GB200 NVL7224,246MI355X36,632 | GB200 NVL7210,580MI355X17,494 |
| Cost ($/M tok) | GB200 NVL72$0.005MI355X$0.005 | GB200 NVL72$0.021MI355X$0.011 | GB200 NVL72$0.049MI355X$0.024 |
| tok/s/MW | GB200 NVL7252,859,865MI355X38,341,359 | GB200 NVL7212,965,951MI355X17,527,503 | GB200 NVL725,657,666MI355X8,370,504 |
| Concurrency | GB200 NVL72~256MI355X~42 | GB200 NVL72~13MI355X~19 | GB200 NVL72~5MI355X~8 |