DeepSeek V4.1 Flash 552B · Chip comparison

DeepSeek V4.1 Flash 552B — GB200 NVL72 vs Vera Rubin NVL72

Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeek V4.1 Flash 552B. 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 →

GB200 NVL72 / Vera Rubin NVL72 on DeepSeek V4.1 Flash 552B at 295 tok/s/user: 523,138 / 2,430,258 tok/s/chip, $0.001 / $0.000 per million tokens. Vera Rubin NVL72 is 58% cheaper per token; Vera Rubin NVL72 delivers 4.6× the tok/s/chip.

Around the middle of the 181–636 tok/s/user interactivity band, at 409 tok/s/user on DeepSeek V4.1 Flash 552B: GB200 NVL72 runs 118,080 tok/s/chip at $0.004 per million tokens, Vera Rubin NVL72 runs 1,708,210 at $0.001. Vera Rubin NVL72 is 87% cheaper per token; Vera Rubin NVL72 delivers 14× the tok/s/chip.

Setting 523 tok/s/user as the target on DeepSeek V4.1 Flash 552B, GB200 NVL72 produces 21,225 tok/s/chip ($0.024 per million tokens) and Vera Rubin NVL72 produces 1,239,284 ($0.001). Vera Rubin NVL72 is 97% cheaper per token; Vera Rubin NVL72 delivers 58× 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.)

View performance per dollar →

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 NVL72523,138Vera Rubin NVL722,430,258
GB200 NVL72118,080Vera Rubin NVL721,708,210
GB200 NVL7221,225Vera Rubin NVL721,239,284
Cost ($/M tok)
GB200 NVL72$0.001Vera Rubin NVL72$0.000
GB200 NVL72$0.004Vera Rubin NVL72$0.001
GB200 NVL72$0.024Vera Rubin NVL72$0.001
tok/s/MW
GB200 NVL72279,753,030Vera Rubin NVL72736,441,769
GB200 NVL7263,144,369Vera Rubin NVL72517,639,343
GB200 NVL7211,350,104Vera Rubin NVL72375,540,482
Concurrency
GB200 NVL72~3,018Vera Rubin NVL72~11,336
GB200 NVL72~816Vera Rubin NVL72~11,016
GB200 NVL72~11Vera Rubin NVL72~7,854