Qwen 3.5 397B-A17B — B300 vs GB200 NVL72 Performance per Dollar
Cost per million tokens of B300 (NVIDIA Blackwell) versus GB200 NVL72 (NVIDIA Blackwell) on Qwen 3.5 397B-A17B. 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.
On Qwen 3.5 397B-A17B at 92 tok/s/user, the per-million math comes out to $0.13 for B300 and $0.05 for GB200 NVL72; GB200 NVL72 delivers 156% more output per dollar.
At 138 tok/s/user on Qwen 3.5 397B-A17B, B300 costs $0.18 per million tokens; GB200 NVL72 costs $0.10. GB200 NVL72 is 82% more cost-efficient at this operating point.
GB200 NVL72 edges B300 at 184 tok/s/user on Qwen 3.5 397B-A17B — $0.15 per million tokens versus $0.25, a 61% cost-per-token gap. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Chip pricing (owning hyperscaler): B300 $2.26/chip/hr · GB200 NVL72 $1.86/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 | B300:$0.129GB200 NVL72:$0.050 | B300:$0.182GB200 NVL72:$0.100 | B300:$0.246GB200 NVL72:$0.153 |
| Concurrency | B300:~25GB200 NVL72:~1042 | B300:~12GB200 NVL72:~133 | B300:~7GB200 NVL72:~41 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.