Qwen 3.5 397B-A17B · Chip comparison
Qwen 3.5 397B-A17B — GB200 NVL72 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of GB200 NVL72 (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.
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 →
Throughput at 191 tok/s/user on Qwen 3.5 397B-A17B: GB200 NVL72 hits 41,390 tok/s/chip, GB300 NVL72 hits 124,058. Per-million costs land at $0.012 and $0.005 respectively. GB300 NVL72 is 141% cheaper per token; GB300 NVL72 delivers 200% more tok/s/chip.
GB200 NVL72 / GB300 NVL72 on Qwen 3.5 397B-A17B at 302 tok/s/user: 22,459 / 87,206 tok/s/chip, $0.023 / $0.007 per million tokens. GB300 NVL72 is 213% cheaper per token; GB300 NVL72 delivers 288% more tok/s/chip.
Toward the upper edge of the 81–523 tok/s/user interactivity band, at 413 tok/s/user on Qwen 3.5 397B-A17B: GB200 NVL72 runs 8,412 tok/s/chip at $0.061/M tokens, GB300 NVL72 runs 46,826 at $0.014/M. GB300 NVL72 is 348% cheaper per token; GB300 NVL72 delivers 457% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
GB300 NVL72 vs GB200 NVL72: full specs & benchmarks across models →
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL7241,389.9GB300 NVL72124,057.9 | GB200 NVL7222,459.3GB300 NVL7287,206.5 | GB200 NVL728,412.5GB300 NVL7246,825.5 |
| Cost ($/M tok) | GB200 NVL72$0.012GB300 NVL72$0.005 | GB200 NVL72$0.023GB300 NVL72$0.007 | GB200 NVL72$0.061GB300 NVL72$0.014 |
| tok/s/MW | GB200 NVL7222,133,631GB300 NVL7258,517,878 | GB200 NVL7212,010,308GB300 NVL7241,135,118 | GB200 NVL724,498,655GB300 NVL7222,087,515 |
| Concurrency | GB200 NVL72~24GB300 NVL72~656 | GB200 NVL72~13GB300 NVL72~225 | GB200 NVL72~5GB300 NVL72~45 |