Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on Qwen 3.8 Flash Next 176B-A6B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β 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 β
Near the low end of the 25β239 tok/s/user interactivity band, at 78 tok/s/user on Qwen 3.8 Flash Next 176B-A6B: GB300 NVL72 runs 83888 tok/s/chip at $0.01/M tokens, Vera Rubin NVL72 runs 219209 at $0.00/M. Vera Rubin NVL72 is 67% cheaper per token; Vera Rubin NVL72 delivers 161% more tok/s/chip.
Setting 132 tok/s/user as the target on Qwen 3.8 Flash Next 176B-A6B, GB300 NVL72 produces 48014 tok/s/chip ($0.01 per million tokens) and Vera Rubin NVL72 produces 189780 ($0.01). Vera Rubin NVL72 is 153% cheaper per token; Vera Rubin NVL72 delivers 295% more tok/s/chip.
At 186 tok/s/user interactivity on Qwen 3.8 Flash Next 176B-A6B, GB300 NVL72 delivers 18703 tok/s/chip at $0.03 per million tokens; Vera Rubin NVL72 delivers 122682 tok/s/chip at $0.01. Vera Rubin NVL72 is 320% cheaper per token; Vera Rubin NVL72 delivers 556% more tok/s/chip at this point. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:83887.7Vera Rubin NVL72:219209.0 | GB300 NVL72:48014.2Vera Rubin NVL72:189779.7 | GB300 NVL72:18703.4Vera Rubin NVL72:122681.8 |
| Cost ($/M tok) | GB300 NVL72:$0.008Vera Rubin NVL72:$0.005 | GB300 NVL72:$0.013Vera Rubin NVL72:$0.005 | GB300 NVL72:$0.034Vera Rubin NVL72:$0.008 |
| tok/s/MW | GB300 NVL72:39569677Vera Rubin NVL72:66426965 | GB300 NVL72:22648213Vera Rubin NVL72:57509009 | GB300 NVL72:8822371Vera Rubin NVL72:37176292 |
| Concurrency | GB300 NVL72:~17Vera Rubin NVL72:~231 | GB300 NVL72:~10Vera Rubin NVL72:~288 | GB300 NVL72:~3Vera Rubin NVL72:~219 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
