Qwen 3.5 397B-A17B · Chip comparison

Qwen 3.5 397B-A17B — B200 vs GB300 NVL72

Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on Qwen 3.5 397B-A17B. 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.

Throughput at 113 tok/s/user on Qwen 3.5 397B-A17B: B200 hits 7828 tok/s/chip, GB300 NVL72 hits 19296. Per-million costs land at $0.06 and $0.03 respectively. GB300 NVL72 is 85% cheaper per token; GB300 NVL72 delivers 146% more tok/s/chip.

B200 / GB300 NVL72 on Qwen 3.5 397B-A17B at 182 tok/s/user: 5087 / 10297 tok/s/chip, $0.10 / $0.06 per million tokens. GB300 NVL72 is 51% cheaper per token; GB300 NVL72 delivers 102% more tok/s/chip.

Toward the upper edge of the 44–320 tok/s/user interactivity band, at 252 tok/s/user on Qwen 3.5 397B-A17B: B200 runs 3039 tok/s/chip at $0.16/M tokens, GB300 NVL72 runs 5122 at $0.13/M. GB300 NVL72 is 28% cheaper per token; GB300 NVL72 delivers 69% more tok/s/chip. (Numbers reflect the default 8k/1k · 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 →

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/chip)
B200:7828.4GB300 NVL72:19295.9
B200:5086.7GB300 NVL72:10296.9
B200:3038.5GB300 NVL72:5122.2
Cost ($/M tok)
B200:$0.061GB300 NVL72:$0.033
B200:$0.095GB300 NVL72:$0.063
B200:$0.163GB300 NVL72:$0.127
tok/s/MW
B200:4578039GB300 NVL72:9101834
B200:2974687GB300 NVL72:4857028
B200:1776909GB300 NVL72:2416149
Concurrency
B200:~21GB300 NVL72:~1030
B200:~7GB300 NVL72:~459
B200:~4GB300 NVL72:~91

Inference Performance

Inference performance metrics across different models, hardware configurations, and serving parameters.

Vendor:
Deployment:
Spec Decoding: