Head-to-head AI inference benchmark comparison of B300 (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 — 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 →
B300 / GB200 NVL72 on DeepSeek V4.1 Flash 552B at 118 tok/s/user: 58064 / 32927 tok/s/chip, $0.01 / $0.02 per million tokens. B300 is 45% cheaper per token; B300 delivers 76% more tok/s/chip.
Around the middle of the 27–392 tok/s/user interactivity band, at 210 tok/s/user on DeepSeek V4.1 Flash 552B: B300 runs 35519 tok/s/chip at $0.02/M tokens, GB200 NVL72 runs 30095 at $0.02/M. GB200 NVL72 is 3% cheaper per token; B300 delivers 18% more tok/s/chip.
Setting 301 tok/s/user as the target on DeepSeek V4.1 Flash 552B, B300 produces 20199 tok/s/chip ($0.03 per million tokens) and GB200 NVL72 produces 18887 ($0.03). GB200 NVL72 is 14% cheaper per token; B300 delivers 7% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:58064.4GB200 NVL72:32926.6 | B300:35519.1GB200 NVL72:30094.8 | B300:20198.9GB200 NVL72:18887.3 |
| Cost ($/M tok) | B300:$0.011GB200 NVL72:$0.016 | B300:$0.018GB200 NVL72:$0.017 | B300:$0.031GB200 NVL72:$0.027 |
| tok/s/MW | B300:30560235GB200 NVL72:17607798 | B300:18694237GB200 NVL72:16093490 | B300:10631017GB200 NVL72:10100150 |
| Concurrency | B300:~42GB200 NVL72:~16 | B300:~10GB200 NVL72:~8 | B300:~5GB200 NVL72:~4 |
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.
