Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and H100 (NVIDIA Hopper) on MiniMax M3.1 Flash 428B. 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
GB300 NVL72 / H100 on MiniMax M3.1 Flash 428B at 84 tok/s/user: 57954 / 5732 tok/s/chip, $0.01 / $0.06 per million tokens. GB300 NVL72 is 412% cheaper per token; GB300 NVL72 delivers 911% more tok/s/chip.
Around the middle of the 17β288 tok/s/user interactivity band, at 152 tok/s/user on MiniMax M3.1 Flash 428B: GB300 NVL72 runs 57954 tok/s/chip at $0.01/M tokens, H100 runs 4486 at $0.07/M. GB300 NVL72 is 554% cheaper per token; GB300 NVL72 delivers 1192% more tok/s/chip.
Setting 220 tok/s/user as the target on MiniMax M3.1 Flash 428B, GB300 NVL72 produces 31494 tok/s/chip ($0.02 per million tokens) and H100 produces 3595 ($0.09). GB300 NVL72 is 344% cheaper per token; GB300 NVL72 delivers 776% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:57953.6H100:5732.4 | GB300 NVL72:57953.6H100:4486.1 | GB300 NVL72:31494.2H100:3594.6 |
| Cost ($/M tok) | GB300 NVL72:$0.011H100:$0.057 | GB300 NVL72:$0.011H100:$0.072 | GB300 NVL72:$0.020H100:$0.090 |
| tok/s/MW | GB300 NVL72:27336609H100:4184237 | GB300 NVL72:27336609H100:3274527 | GB300 NVL72:14855771H100:2623785 |
| Concurrency | GB300 NVL72:~32H100:~9 | GB300 NVL72:~32H100:~7 | GB300 NVL72:~18H100:~3 |
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.
