Kimi K3.1 Pro — B300 vs GB200 NVL72
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and GB200 NVL72 (NVIDIA Blackwell) on Kimi K3.1 Pro. 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 →
Throughput at 73 tok/s/user on Kimi K3.1 Pro: B300 hits 7627 tok/s/chip, GB200 NVL72 hits 8199. Per-million costs land at $0.08 and $0.06 respectively. GB200 NVL72 is 31% cheaper per token; GB200 NVL72 delivers 8% more tok/s/chip.
B300 / GB200 NVL72 on Kimi K3.1 Pro at 119 tok/s/user: 5683 / 3877 tok/s/chip, $0.11 / $0.13 per million tokens. B300 is 21% cheaper per token; B300 delivers 47% more tok/s/chip.
Toward the upper edge of the 27–211 tok/s/user interactivity band, at 165 tok/s/user on Kimi K3.1 Pro: B300 runs 4371 tok/s/chip at $0.14/M tokens, GB200 NVL72 runs 1787 at $0.29/M. B300 is 101% cheaper per token; B300 delivers 145% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:7626.9GB200 NVL72:8199.4 | B300:5683.1GB200 NVL72:3877.2 | B300:4370.9GB200 NVL72:1786.7 |
| Cost ($/M tok) | B300:$0.082GB200 NVL72:$0.063 | B300:$0.110GB200 NVL72:$0.133 | B300:$0.144GB200 NVL72:$0.289 |
| tok/s/MW | B300:4014153GB200 NVL72:4384723 | B300:2991089GB200 NVL72:2073380 | B300:2300493GB200 NVL72:955429 |
| Concurrency | B300:~16GB200 NVL72:~46 | B300:~8GB200 NVL72:~20 | B300:~6GB200 NVL72:~5 |