Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and GB200 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 8–198 tok/s/user interactivity band, at 56 tok/s/user on Kimi K3 2.8T: B300 runs 5680 tok/s/chip at $0.11/M tokens, GB200 NVL72 runs 4090 at $0.13/M. B300 is 14% cheaper per token; B300 delivers 39% more tok/s/chip.
Setting 103 tok/s/user as the target on Kimi K3 2.8T, B300 produces 3971 tok/s/chip ($0.16 per million tokens) and GB200 NVL72 produces 2300 ($0.22). B300 is 42% cheaper per token; B300 delivers 73% more tok/s/chip.
At 151 tok/s/user interactivity on Kimi K3 2.8T, B300 delivers 2073 tok/s/chip at $0.30 per million tokens; GB200 NVL72 delivers 1020 tok/s/chip at $0.51. B300 is 67% cheaper per token; B300 delivers 103% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:5679.5GB200 NVL72:4089.6 | B300:3971.0GB200 NVL72:2299.8 | B300:2073.4GB200 NVL72:1019.9 |
| Cost ($/M tok) | B300:$0.111GB200 NVL72:$0.126 | B300:$0.158GB200 NVL72:$0.225 | B300:$0.303GB200 NVL72:$0.507 |
| tok/s/MW | B300:2989225GB200 NVL72:2186971 | B300:2090009GB200 NVL72:1229827 | B300:1091265GB200 NVL72:545383 |
| Concurrency | B300:~9GB200 NVL72:~16 | B300:~6GB200 NVL72:~16 | B300:~2GB200 NVL72:~2 |
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.
