Kimi K3.1 Pro — GB200 NVL72 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 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 →
At 99 tok/s/user interactivity on Kimi K3.1 Pro, GB200 NVL72 delivers 5916 tok/s/chip at $0.09 per million tokens; GB300 NVL72 delivers 16725 tok/s/chip at $0.04. GB300 NVL72 is 128% cheaper per token; GB300 NVL72 delivers 183% more tok/s/chip at this point.
GB200 NVL72 posts 2629 tok/s/chip for $0.20 per million tokens at 137 tok/s/user on Kimi K3.1 Pro; GB300 NVL72 posts 10838 tok/s/chip for $0.06. GB300 NVL72 is 232% cheaper per token; GB300 NVL72 delivers 312% more tok/s/chip.
Throughput at 174 tok/s/user on Kimi K3.1 Pro: GB200 NVL72 hits 1570 tok/s/chip, GB300 NVL72 hits 6970. Per-million costs land at $0.33 and $0.09 respectively. GB300 NVL72 is 258% cheaper per token; GB300 NVL72 delivers 344% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
View performance-per-dollar view →
GB300 NVL72 vs GB200 NVL72: full specs & benchmarks across models →
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:5916.3GB300 NVL72:16724.9 | GB200 NVL72:2629.3GB300 NVL72:10838.3 | GB200 NVL72:1569.7GB300 NVL72:6970.4 |
| Cost ($/M tok) | GB200 NVL72:$0.087GB300 NVL72:$0.038 | GB200 NVL72:$0.197GB300 NVL72:$0.059 | GB200 NVL72:$0.329GB300 NVL72:$0.092 |
| tok/s/MW | GB200 NVL72:3163796GB300 NVL72:7889109 | GB200 NVL72:1406046GB300 NVL72:5112411 | GB200 NVL72:839404GB300 NVL72:3287928 |
| Concurrency | GB200 NVL72:~32GB300 NVL72:~75 | GB200 NVL72:~13GB300 NVL72:~55 | GB200 NVL72:~3GB300 NVL72:~34 |