Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and B300 (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 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
Throughput at 53 tok/s/user on Kimi K3 2.8T: B200 hits 5834 tok/s/chip, B300 hits 5877. Per-million costs land at $0.08 and $0.11 respectively. B200 is 30% cheaper per token; throughput per chip is essentially tied.
B200 / B300 on Kimi K3 2.8T at 98 tok/s/user: 2754 / 4098 tok/s/chip, $0.17 / $0.15 per million tokens. B300 is 14% cheaper per token; B300 delivers 49% more tok/s/chip.
Toward the upper edge of the 9β186 tok/s/user interactivity band, at 142 tok/s/user on Kimi K3 2.8T: B200 runs 1781 tok/s/chip at $0.27/M tokens, B300 runs 2089 at $0.30/M. B200 is 11% cheaper per token; B300 delivers 17% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:5833.9B300:5876.9 | B200:2753.6B300:4098.3 | B200:1781.1B300:2089.2 |
| Cost ($/M tok) | B200:$0.082B300:$0.107 | B200:$0.175B300:$0.153 | B200:$0.270B300:$0.300 |
| tok/s/MW | B200:3411616B300:3093098 | B200:1610277B300:2157004 | B200:1041568B300:1099598 |
| Concurrency | B200:~19B300:~10 | B200:~8B300:~6 | B200:~5B300:~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.
