MiniMax M3.1 Flash 428B · Chip comparison
MiniMax M3.1 Flash 428B — B300 vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) 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 442 tok/s/user as the target on MiniMax M3.1 Flash 428B, B300 produces 13,225 tok/s/chip ($0.047 per million tokens) and Vera Rubin NVL72 produces 157,300 ($0.006). Vera Rubin NVL72 is 87% cheaper per token; Vera Rubin NVL72 delivers 12× the tok/s/chip.
At 508 tok/s/user interactivity on MiniMax M3.1 Flash 428B, B300 delivers 9,055 tok/s/chip at $0.069 per million tokens; Vera Rubin NVL72 delivers 86,029 tok/s/chip at $0.012. Vera Rubin NVL72 is 83% cheaper per token; Vera Rubin NVL72 delivers 9.5× the tok/s/chip at this point.
B300 posts 8,036 tok/s/chip for $0.078 per million tokens at 575 tok/s/user on MiniMax M3.1 Flash 428B; Vera Rubin NVL72 posts 68,202 tok/s/chip for $0.015. Vera Rubin NVL72 is 81% cheaper per token; Vera Rubin NVL72 delivers 8.5× the tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B30013,225Vera Rubin NVL72157,300 | B3009,055Vera Rubin NVL7286,029 | B3008,036Vera Rubin NVL7268,202 |
| Cost ($/M tok) | B300$0.047Vera Rubin NVL72$0.006 | B300$0.069Vera Rubin NVL72$0.012 | B300$0.078Vera Rubin NVL72$0.015 |
| tok/s/MW | B3006,960,741Vera Rubin NVL7247,666,736 | B3004,765,985Vera Rubin NVL7226,069,261 | B3004,229,668Vera Rubin NVL7220,667,336 |
| Concurrency | B300~7Vera Rubin NVL72~208 | B300~3Vera Rubin NVL72~128 | B300~2Vera Rubin NVL72~81 |