Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeekv4 Pro 0813 1.6T. 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 posts 12019 tok/s/chip for $0.05 per million tokens at 125 tok/s/user on DeepSeekv4 Pro 0813 1.6T; Vera Rubin NVL72 posts 170118 tok/s/chip for $0.01. Vera Rubin NVL72 is 786% cheaper per token; Vera Rubin NVL72 delivers 1315% more tok/s/chip.
Throughput at 175 tok/s/user on DeepSeekv4 Pro 0813 1.6T: B300 hits 5273 tok/s/chip, Vera Rubin NVL72 hits 61937. Per-million costs land at $0.12 and $0.02 respectively. Vera Rubin NVL72 is 635% cheaper per token; Vera Rubin NVL72 delivers 1075% more tok/s/chip.
B300 / Vera Rubin NVL72 on DeepSeekv4 Pro 0813 1.6T at 226 tok/s/user: 3778 / 7223 tok/s/chip, $0.17 / $0.14 per million tokens. Vera Rubin NVL72 is 20% cheaper per token; Vera Rubin NVL72 delivers 91% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:12019.3Vera Rubin NVL72:170117.7 | B300:5272.9Vera Rubin NVL72:61936.7 | B300:3778.2Vera Rubin NVL72:7223.3 |
| Cost ($/M tok) | B300:$0.052Vera Rubin NVL72:$0.006 | B300:$0.119Vera Rubin NVL72:$0.016 | B300:$0.166Vera Rubin NVL72:$0.139 |
| tok/s/MW | B300:6325931Vera Rubin NVL72:51550815 | B300:2775206Vera Rubin NVL72:18768687 | B300:1988518Vera Rubin NVL72:2188894 |
| Concurrency | B300:~8Vera Rubin NVL72:~1178 | B300:~7Vera Rubin NVL72:~547 | B300:~4Vera Rubin NVL72:~33 |
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
DeepSeek V4 Pro 0813 1.6T 8K / 1K Total Tokens per $1 TCO vs. Interactivity