Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on Kimi K3 2.8T. 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 →
Near the low end of the 19–205 tok/s/user interactivity band, at 65 tok/s/user on Kimi K3 2.8T: B200 runs 3600 tok/s/chip at $0.13/M tokens, GB300 NVL72 runs 13116 at $0.05/M. GB300 NVL72 is 173% cheaper per token; GB300 NVL72 delivers 264% more tok/s/chip.
Setting 112 tok/s/user as the target on Kimi K3 2.8T, B200 produces 1916 tok/s/chip ($0.25 per million tokens) and GB300 NVL72 produces 8929 ($0.07). GB300 NVL72 is 249% cheaper per token; GB300 NVL72 delivers 366% more tok/s/chip.
At 159 tok/s/user interactivity on Kimi K3 2.8T, B200 delivers 1235 tok/s/chip at $0.39 per million tokens; GB300 NVL72 delivers 5937 tok/s/chip at $0.11. GB300 NVL72 is 260% cheaper per token; GB300 NVL72 delivers 381% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:3600.3GB300 NVL72:13116.1 | B200:1916.0GB300 NVL72:8928.8 | B200:1234.7GB300 NVL72:5937.4 |
| Cost ($/M tok) | B200:$0.133GB300 NVL72:$0.049 | B200:$0.251GB300 NVL72:$0.072 | B200:$0.389GB300 NVL72:$0.108 |
| tok/s/MW | B200:2105461GB300 NVL72:6186846 | B200:1120449GB300 NVL72:4211685 | B200:722045GB300 NVL72:2800680 |
| Concurrency | B200:~12GB300 NVL72:~56 | B200:~5GB300 NVL72:~48 | B200:~2GB300 NVL72:~27 |
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.
