Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) on DeepSeek R1. 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.
Setting 57 tok/s/user as the target on DeepSeek R1, GB300 NVL72 produces 8552 tok/s/chip ($0.08 per million tokens) and H200 produces 1302 ($0.26). GB300 NVL72 is 247% cheaper per token; GB300 NVL72 delivers 557% more tok/s/chip.
At 94 tok/s/user interactivity on DeepSeek R1, GB300 NVL72 delivers 5667 tok/s/chip at $0.11 per million tokens; H200 delivers 779 tok/s/chip at $0.43. GB300 NVL72 is 284% cheaper per token; GB300 NVL72 delivers 627% more tok/s/chip at this point.
GB300 NVL72 posts 3404 tok/s/chip for $0.19 per million tokens at 130 tok/s/user on DeepSeek R1; H200 posts 611 tok/s/chip for $0.55. GB300 NVL72 is 194% cheaper per token; GB300 NVL72 delivers 457% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:8552.1H200:1302.1 | GB300 NVL72:5667.4H200:779.2 | GB300 NVL72:3404.3H200:611.1 |
| Cost ($/M tok) | GB300 NVL72:$0.075H200:$0.260 | GB300 NVL72:$0.113H200:$0.435 | GB300 NVL72:$0.188H200:$0.555 |
| tok/s/MW | GB300 NVL72:4033997H200:950421 | GB300 NVL72:2673310H200:568790 | GB300 NVL72:1605781H200:446083 |
| Concurrency | GB300 NVL72:~1226H200:~109 | GB300 NVL72:~456H200:~8 | GB300 NVL72:~222H200:~4 |
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 R1 0528 671B 8K / 1K Total Tokens per $1 TCO vs. Interactivity