Head-to-head AI inference benchmark comparison of B300 (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 →
B300 / GB300 NVL72 on Qwen 3.5 397B-A17B at 144 tok/s/user: 30673 / 58759 tok/s/chip, $0.02 / $0.01 per million tokens. GB300 NVL72 is 87% cheaper per token; GB300 NVL72 delivers 92% more tok/s/chip.
Around the middle of the 68–374 tok/s/user interactivity band, at 221 tok/s/user on Qwen 3.5 397B-A17B: B300 runs 19493 tok/s/chip at $0.03/M tokens, GB300 NVL72 runs 32170 at $0.02/M. GB300 NVL72 is 61% cheaper per token; GB300 NVL72 delivers 65% more tok/s/chip.
Setting 298 tok/s/user as the target on Qwen 3.5 397B-A17B, B300 produces 5525 tok/s/chip ($0.11 per million tokens) and GB300 NVL72 produces 24026 ($0.03). GB300 NVL72 is 325% cheaper per token; GB300 NVL72 delivers 335% more tok/s/chip. (Numbers reflect the default agentic-traces · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:30673.0GB300 NVL72:58759.3 | B300:19493.5GB300 NVL72:32170.4 | B300:5525.3GB300 NVL72:24026.3 |
| Cost ($/M tok) | B300:$0.020GB300 NVL72:$0.011 | B300:$0.032GB300 NVL72:$0.020 | B300:$0.114GB300 NVL72:$0.027 |
| tok/s/MW | B300:16143683GB300 NVL72:27716671 | B300:10259718GB300 NVL72:15174719 | B300:2908066GB300 NVL72:11333178 |
| Concurrency | B300:~22GB300 NVL72:~380 | B300:~30GB300 NVL72:~153 | B300:~4GB300 NVL72:~109 |
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