Qwen 3.5 397B-A17B — B300 vs RTX PRO 6000
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and RTX PRO 6000 (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.
B300 posts 14511 tok/s/chip for $0.04 per million tokens at 59 tok/s/user on Qwen 3.5 397B-A17B; RTX PRO 6000 posts 947 tok/s/chip for $0.19. B300 is 335% cheaper per token; B300 delivers 1432% more tok/s/chip.
Throughput at 91 tok/s/user on Qwen 3.5 397B-A17B: B300 hits 11105 tok/s/chip, RTX PRO 6000 hits 734. Per-million costs land at $0.06 and $0.26 respectively. B300 is 355% cheaper per token; B300 delivers 1413% more tok/s/chip.
B300 / RTX PRO 6000 on Qwen 3.5 397B-A17B at 122 tok/s/user: 8765 / 511 tok/s/chip, $0.07 / $0.41 per million tokens. B300 is 467% cheaper per token; B300 delivers 1615% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:14510.6RTX PRO 6000:947.2 | B300:11104.7RTX PRO 6000:734.1 | B300:8764.9RTX PRO 6000:511.2 |
| Cost ($/M tok) | B300:$0.043RTX PRO 6000:$0.188 | B300:$0.057RTX PRO 6000:$0.257 | B300:$0.072RTX PRO 6000:$0.406 |
| tok/s/MW | B300:7637143RTX PRO 6000:971508 | B300:5844587RTX PRO 6000:752914 | B300:4613083RTX PRO 6000:524318 |
| Concurrency | B300:~60RTX PRO 6000:~9 | B300:~30RTX PRO 6000:~4 | B300:~18RTX PRO 6000:~2 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.