MiniMax M3.1 Flash 428B · Chip comparison
MiniMax M3.1 Flash 428B — MI355X vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of MI355X (AMD CDNA 4) and Vera Rubin NVL72 (NVIDIA Vera Rubin) 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 →
Setting 430 tok/s/user as the target on MiniMax M3.1 Flash 428B, MI355X produces 9,678 tok/s/chip ($0.043 per million tokens) and Vera Rubin NVL72 produces 173,899 ($0.006). Vera Rubin NVL72 is 87% cheaper per token; Vera Rubin NVL72 delivers 18× the tok/s/chip.
At 484 tok/s/user interactivity on MiniMax M3.1 Flash 428B, MI355X delivers 8,610 tok/s/chip at $0.048 per million tokens; Vera Rubin NVL72 delivers 103,421 tok/s/chip at $0.010. Vera Rubin NVL72 is 80% cheaper per token; Vera Rubin NVL72 delivers 12× the tok/s/chip at this point.
MI355X posts 7,999 tok/s/chip for $0.052 per million tokens at 538 tok/s/user on MiniMax M3.1 Flash 428B; Vera Rubin NVL72 posts 77,140 tok/s/chip for $0.013. Vera Rubin NVL72 is 75% cheaper per token; Vera Rubin NVL72 delivers 9.6× the tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI355X9,678Vera Rubin NVL72173,899 | MI355X8,610Vera Rubin NVL72103,421 | MI355X7,999Vera Rubin NVL7277,140 |
| Cost ($/M tok) | MI355X$0.043Vera Rubin NVL72$0.006 | MI355X$0.048Vera Rubin NVL72$0.010 | MI355X$0.052Vera Rubin NVL72$0.013 |
| tok/s/MW | MI355X4,630,395Vera Rubin NVL7252,696,630 | MI355X4,119,498Vera Rubin NVL7231,339,659 | MI355X3,827,134Vera Rubin NVL7223,375,630 |
| Concurrency | MI355X~4Vera Rubin NVL72~238 | MI355X~3Vera Rubin NVL72~147 | MI355X~2Vera Rubin NVL72~105 |