Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on DeepSeekv4 Pro 0813 1.6T. 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 →
Near the low end of the 53–285 tok/s/user interactivity band, at 111 tok/s/user on DeepSeekv4 Pro 0813 1.6T: GB200 NVL72 runs 30098 tok/s/chip at $0.02/M tokens, GB300 NVL72 runs 30421 at $0.02/M. GB200 NVL72 is 23% cheaper per token; GB300 NVL72 delivers 1% more tok/s/chip.
Setting 169 tok/s/user as the target on DeepSeekv4 Pro 0813 1.6T, GB200 NVL72 produces 13108 tok/s/chip ($0.04 per million tokens) and GB300 NVL72 produces 8949 ($0.07). GB200 NVL72 is 82% cheaper per token; GB200 NVL72 delivers 46% more tok/s/chip.
At 227 tok/s/user interactivity on DeepSeekv4 Pro 0813 1.6T, GB200 NVL72 delivers 4252 tok/s/chip at $0.12 per million tokens; GB300 NVL72 delivers 4517 tok/s/chip at $0.14. GB200 NVL72 is 17% cheaper per token; GB300 NVL72 delivers 6% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:30098.0GB300 NVL72:30420.6 | GB200 NVL72:13108.1GB300 NVL72:8948.5 | GB200 NVL72:4252.2GB300 NVL72:4516.9 |
| Cost ($/M tok) | GB200 NVL72:$0.017GB300 NVL72:$0.021 | GB200 NVL72:$0.039GB300 NVL72:$0.072 | GB200 NVL72:$0.122GB300 NVL72:$0.142 |
| tok/s/MW | GB200 NVL72:16095213GB300 NVL72:14349324 | GB200 NVL72:7009654GB300 NVL72:4221003 | GB200 NVL72:2273893GB300 NVL72:2130613 |
| Concurrency | GB200 NVL72:~330GB300 NVL72:~250 | GB200 NVL72:~43GB300 NVL72:~7 | GB200 NVL72:~5GB300 NVL72:~5 |
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 V4 Pro 0813 1.6T 8K / 1K Total Tokens per $1 TCO vs. Interactivity