Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 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 — 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 118 tok/s/user as the target on DeepSeek V4.1 Flash 552B, GB200 NVL72 produces 32927 tok/s/chip ($0.02 per million tokens) and GB300 NVL72 produces 66547 ($0.01). GB300 NVL72 is 63% cheaper per token; GB300 NVL72 delivers 102% more tok/s/chip.
At 221 tok/s/user interactivity on DeepSeek V4.1 Flash 552B, GB200 NVL72 delivers 29300 tok/s/chip at $0.02 per million tokens; GB300 NVL72 delivers 29451 tok/s/chip at $0.02. GB200 NVL72 is 24% cheaper per token; throughput per chip is essentially tied at this point.
GB200 NVL72 posts 14949 tok/s/chip for $0.03 per million tokens at 323 tok/s/user on DeepSeek V4.1 Flash 552B; GB300 NVL72 posts 15706 tok/s/chip for $0.04. GB200 NVL72 is 18% cheaper per token; GB300 NVL72 delivers 5% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:32926.6GB300 NVL72:66547.5 | GB200 NVL72:29300.4GB300 NVL72:29451.1 | GB200 NVL72:14949.5GB300 NVL72:15705.9 |
| Cost ($/M tok) | GB200 NVL72:$0.016GB300 NVL72:$0.010 | GB200 NVL72:$0.018GB300 NVL72:$0.022 | GB200 NVL72:$0.035GB300 NVL72:$0.041 |
| tok/s/MW | GB200 NVL72:17607798GB300 NVL72:31390310 | GB200 NVL72:15668663GB300 NVL72:13892014 | GB200 NVL72:7994379GB300 NVL72:7408429 |
| Concurrency | GB200 NVL72:~16GB300 NVL72:~21 | GB200 NVL72:~7GB300 NVL72:~8 | GB200 NVL72:~3GB300 NVL72:~3 |
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.
