Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB200 NVL72 (NVIDIA Blackwell) on DeepSeek V4.1 Flash 552B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β 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 β
Throughput at 166 tok/s/user on DeepSeek V4.1 Flash 552B: B200 hits 116628 tok/s/chip, GB200 NVL72 hits 90644. Per-million costs land at $0.00 and $0.01 respectively. B200 is 38% cheaper per token; B200 delivers 29% more tok/s/chip.
B200 / GB200 NVL72 on DeepSeek V4.1 Flash 552B at 282 tok/s/user: 31925 / 16669 tok/s/chip, $0.02 / $0.03 per million tokens. B200 is 106% cheaper per token; B200 delivers 92% more tok/s/chip.
Toward the upper edge of the 50β515 tok/s/user interactivity band, at 399 tok/s/user on DeepSeek V4.1 Flash 552B: B200 runs 16060 tok/s/chip at $0.03/M tokens, GB200 NVL72 runs 15930 at $0.03/M. B200 is 8% cheaper per token; throughput per chip is essentially tied. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:116627.9GB200 NVL72:90643.8 | B200:31925.4GB200 NVL72:16669.4 | B200:16060.4GB200 NVL72:15930.4 |
| Cost ($/M tok) | B200:$0.004GB200 NVL72:$0.006 | B200:$0.015GB200 NVL72:$0.031 | B200:$0.030GB200 NVL72:$0.032 |
| tok/s/MW | B200:68203453GB200 NVL72:48472624 | B200:18669812GB200 NVL72:8914122 | B200:9392070GB200 NVL72:8518914 |
| Concurrency | B200:~47GB200 NVL72:~47 | B200:~10GB200 NVL72:~4 | B200:~8GB200 NVL72:~8 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
