Muse Spark Open 816B — GB300 NVL72 vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on Muse Spark Open 816B. 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 →
GB300 NVL72 / Vera Rubin NVL72 on Muse Spark Open 816B at 259 tok/s/user: 18538 / 64658 tok/s/chip, $0.03 / $0.02 per million tokens. Vera Rubin NVL72 is 123% cheaper per token; Vera Rubin NVL72 delivers 249% more tok/s/chip.
Around the middle of the 107–714 tok/s/user interactivity band, at 411 tok/s/user on Muse Spark Open 816B: GB300 NVL72 runs 8379 tok/s/chip at $0.08/M tokens, Vera Rubin NVL72 runs 41314 at $0.02/M. Vera Rubin NVL72 is 215% cheaper per token; Vera Rubin NVL72 delivers 393% more tok/s/chip.
Setting 562 tok/s/user as the target on Muse Spark Open 816B, GB300 NVL72 produces 4575 tok/s/chip ($0.14 per million tokens) and Vera Rubin NVL72 produces 27187 ($0.04). Vera Rubin NVL72 is 280% cheaper per token; Vera Rubin NVL72 delivers 494% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:18538.1Vera Rubin NVL72:64657.6 | GB300 NVL72:8379.2Vera Rubin NVL72:41313.9 | GB300 NVL72:4575.3Vera Rubin NVL72:27187.5 |
| Cost ($/M tok) | GB300 NVL72:$0.035Vera Rubin NVL72:$0.016 | GB300 NVL72:$0.077Vera Rubin NVL72:$0.024 | GB300 NVL72:$0.140Vera Rubin NVL72:$0.037 |
| tok/s/MW | GB300 NVL72:8744368Vera Rubin NVL72:19593211 | GB300 NVL72:3952447Vera Rubin NVL72:12519367 | GB300 NVL72:2158147Vera Rubin NVL72:8238636 |
| Concurrency | GB300 NVL72:~12Vera Rubin NVL72:~35 | GB300 NVL72:~9Vera Rubin NVL72:~39 | GB300 NVL72:~1Vera Rubin NVL72:~100 |