Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and GB200 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 →
At 127 tok/s/user interactivity on Qwen 3.5 397B-A17B, B300 delivers 46569 tok/s/chip at $0.01 per million tokens; GB200 NVL72 delivers 68282 tok/s/chip at $0.01. GB200 NVL72 is 78% cheaper per token; GB200 NVL72 delivers 47% more tok/s/chip at this point.
B300 posts 31336 tok/s/chip for $0.02 per million tokens at 211 tok/s/user on Qwen 3.5 397B-A17B; GB200 NVL72 posts 35151 tok/s/chip for $0.01. GB200 NVL72 is 36% cheaper per token; GB200 NVL72 delivers 12% more tok/s/chip.
Throughput at 295 tok/s/user on Qwen 3.5 397B-A17B: B300 hits 12039 tok/s/chip, GB200 NVL72 hits 22698. Per-million costs land at $0.05 and $0.02 respectively. GB200 NVL72 is 129% cheaper per token; GB200 NVL72 delivers 89% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:46569.0GB200 NVL72:68282.4 | B300:31335.7GB200 NVL72:35151.4 | B300:12039.4GB200 NVL72:22698.5 |
| Cost ($/M tok) | B300:$0.013GB200 NVL72:$0.008 | B300:$0.020GB200 NVL72:$0.015 | B300:$0.052GB200 NVL72:$0.023 |
| tok/s/MW | B300:24510011GB200 NVL72:36514635 | B300:16492458GB200 NVL72:18797560 | B300:6336537GB200 NVL72:12138235 |
| Concurrency | B300:~38GB200 NVL72:~23 | B300:~21GB200 NVL72:~24 | B300:~5GB200 NVL72:~13 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
Qwen3.5 397B 8K / 1K Total Tokens per $1 TCO vs. Interactivity