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 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
B300 / GB200 NVL72 on DeepSeek V4.1 Flash 552B at 224 tok/s/user: 87187 / 61704 tok/s/chip, $0.01 / $0.01 per million tokens. B300 is 16% cheaper per token; B300 delivers 41% more tok/s/chip.
Around the middle of the 84β645 tok/s/user interactivity band, at 364 tok/s/user on DeepSeek V4.1 Flash 552B: B300 runs 39152 tok/s/chip at $0.02/M tokens, GB200 NVL72 runs 38365 at $0.01/M. GB200 NVL72 is 19% cheaper per token; B300 delivers 2% more tok/s/chip.
Setting 505 tok/s/user as the target on DeepSeek V4.1 Flash 552B, B300 produces 27418 tok/s/chip ($0.02 per million tokens) and GB200 NVL72 produces 12062 ($0.04). B300 is 87% cheaper per token; B300 delivers 127% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:87187.0GB200 NVL72:61703.9 | B300:39151.8GB200 NVL72:38364.9 | B300:27417.6GB200 NVL72:12062.0 |
| Cost ($/M tok) | B300:$0.007GB200 NVL72:$0.008 | B300:$0.016GB200 NVL72:$0.013 | B300:$0.023GB200 NVL72:$0.043 |
| tok/s/MW | B300:45887895GB200 NVL72:32996714 | B300:20606231GB200 NVL72:20515982 | B300:14430299GB200 NVL72:6450265 |
| Concurrency | B300:~29GB200 NVL72:~32 | B300:~20GB200 NVL72:~22 | B300:~14GB200 NVL72:~5 |
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.
