Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on GLM 5.3 744B. 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 →
GB200 NVL72 / GB300 NVL72 on GLM 5.3 744B at 143 tok/s/user: 12412 / 14337 tok/s/chip, $0.04 / $0.04 per million tokens. GB200 NVL72 is 8% cheaper per token; GB300 NVL72 delivers 16% more tok/s/chip.
Around the middle of the 92–297 tok/s/user interactivity band, at 195 tok/s/user on GLM 5.3 744B: GB200 NVL72 runs 8529 tok/s/chip at $0.06/M tokens, GB300 NVL72 runs 8991 at $0.07/M. GB200 NVL72 is 18% cheaper per token; GB300 NVL72 delivers 5% more tok/s/chip.
Setting 246 tok/s/user as the target on GLM 5.3 744B, GB200 NVL72 produces 4979 tok/s/chip ($0.10 per million tokens) and GB300 NVL72 produces 6250 ($0.10). GB300 NVL72 is 1% cheaper per token; GB300 NVL72 delivers 26% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:12412.1GB300 NVL72:14337.4 | GB200 NVL72:8528.6GB300 NVL72:8990.8 | GB200 NVL72:4978.7GB300 NVL72:6250.0 |
| Cost ($/M tok) | GB200 NVL72:$0.042GB300 NVL72:$0.045 | GB200 NVL72:$0.061GB300 NVL72:$0.071 | GB200 NVL72:$0.104GB300 NVL72:$0.103 |
| tok/s/MW | GB200 NVL72:6637513GB300 NVL72:6762913 | GB200 NVL72:4560734GB300 NVL72:4240924 | GB200 NVL72:2662432GB300 NVL72:2948110 |
| Concurrency | GB200 NVL72:~63GB300 NVL72:~56 | GB200 NVL72:~39GB300 NVL72:~22 | GB200 NVL72:~5GB300 NVL72:~15 |
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.
