Kimi K3 2.8T — B200 vs GB300 NVL72 Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus GB300 NVL72 (NVIDIA Blackwell) on Kimi K3 2.8T. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. 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: $0.12 per million tokens. GB300 NVL72: $0.08. Both at 57 tok/s/user on Kimi K3 2.8T, with GB300 NVL72 46% cheaper.
Around the middle of the 7–207 tok/s/user interactivity band — at 107 tok/s/user — B200 runs $0.24 per million tokens on Kimi K3 2.8T while GB300 NVL72 runs $0.13. GB300 NVL72 is the cheaper choice by 89%.
On Kimi K3 2.8T at 157 tok/s/user, the per-million math comes out to $0.37 for B200 and $0.19 for GB300 NVL72; GB300 NVL72 delivers 92% more output per dollar. (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.)
Chip pricing (owning hyperscaler): B200 $1.73/chip/hr · GB300 NVL72 $2.31/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.

| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.120GB300 NVL72:$0.082 | B200:$0.241GB300 NVL72:$0.127 | B200:$0.373GB300 NVL72:$0.194 |
| Concurrency | B200:~14GB300 NVL72:~63 | B200:~6GB300 NVL72:~34 | B200:~3GB300 NVL72:~7 |