Head-to-head AI inference benchmark comparison of B200 (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 120 tok/s/user as the target on DeepSeek V4.1 Flash 552B, B200 produces 58921 tok/s/chip ($0.01 per million tokens) and GB300 NVL72 produces 61011 ($0.01). B200 is 29% cheaper per token; GB300 NVL72 delivers 4% more tok/s/chip.
At 223 tok/s/user interactivity on DeepSeek V4.1 Flash 552B, B200 delivers 32400 tok/s/chip at $0.01 per million tokens; GB300 NVL72 delivers 29334 tok/s/chip at $0.02. B200 is 47% cheaper per token; B200 delivers 10% more tok/s/chip at this point.
B200 posts 16566 tok/s/chip for $0.03 per million tokens at 327 tok/s/user on DeepSeek V4.1 Flash 552B; GB300 NVL72 posts 14792 tok/s/chip for $0.04. B200 is 50% cheaper per token; B200 delivers 12% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:58921.1GB300 NVL72:61010.8 | B200:32400.0GB300 NVL72:29334.4 | B200:16565.6GB300 NVL72:14791.9 |
| Cost ($/M tok) | B200:$0.008GB300 NVL72:$0.011 | B200:$0.015GB300 NVL72:$0.022 | B200:$0.029GB300 NVL72:$0.043 |
| tok/s/MW | B200:34456760GB300 NVL72:28778661 | B200:18947340GB300 NVL72:13836972 | B200:9687502GB300 NVL72:6977306 |
| Concurrency | B200:~38GB300 NVL72:~18 | B200:~9GB300 NVL72:~8 | B200:~7GB300 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.
