Qwen 3.5 397B-A17B — B200 vs GB200 NVL72 Performance per Dollar
Cost per million tokens of B200 (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.
At 95 tok/s/user on Qwen 3.5 397B-A17B, B200 costs $0.11 per million tokens; GB200 NVL72 costs $0.06. GB200 NVL72 is 89% more cost-efficient at this operating point.
GB200 NVL72 edges B200 at 154 tok/s/user on Qwen 3.5 397B-A17B — $0.15 per million tokens versus $0.15, a 2% cost-per-token gap.
Push Qwen 3.5 397B-A17B to 214 tok/s/user and B200 lands at $0.21 per million tokens against GB200 NVL72's $0.32 — B200 pulls ahead by 50%. (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): B200 $1.73/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 | B200:$0.107GB200 NVL72:$0.057 | B200:$0.154GB200 NVL72:$0.150 | B200:$0.214GB200 NVL72:$0.320 |
| Concurrency | B200:~23GB200 NVL72:~678 | B200:~10GB200 NVL72:~54 | B200:~5GB200 NVL72:~11 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.