Kimi K3 2.8T — B200 vs GB200 NVL72
Head-to-head AI inference benchmark comparison of B200 (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 →
B200 posts 4595 tok/s/chip for $0.10 per million tokens at 65 tok/s/user on Kimi K3 2.8T; GB200 NVL72 posts 7752 tok/s/chip for $0.07. GB200 NVL72 is 57% cheaper per token; GB200 NVL72 delivers 69% more tok/s/chip.
Throughput at 106 tok/s/user on Kimi K3 2.8T: B200 hits 2477 tok/s/chip, GB200 NVL72 hits 4812. Per-million costs land at $0.19 and $0.11 respectively. GB200 NVL72 is 81% cheaper per token; GB200 NVL72 delivers 94% more tok/s/chip.
B200 / GB200 NVL72 on Kimi K3 2.8T at 146 tok/s/user: 1719 / 3112 tok/s/chip, $0.28 / $0.17 per million tokens. GB200 NVL72 is 68% cheaper per token; GB200 NVL72 delivers 81% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
View performance-per-dollar view →
GB200 NVL72 vs B200: full specs & benchmarks across models →
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:4594.7GB200 NVL72:7751.7 | B200:2476.5GB200 NVL72:4812.2 | B200:1719.2GB200 NVL72:3111.9 |
| Cost ($/M tok) | B200:$0.105GB200 NVL72:$0.067 | B200:$0.194GB200 NVL72:$0.107 | B200:$0.280GB200 NVL72:$0.166 |
| tok/s/MW | B200:2686936GB200 NVL72:4145303 | B200:1448251GB200 NVL72:2573343 | B200:1005383GB200 NVL72:1664096 |
| Concurrency | B200:~15GB200 NVL72:~30 | B200:~7GB200 NVL72:~24 | B200:~5GB200 NVL72:~16 |