DeepSeekv4 Pro 0813 1.6T — B200 vs GB200 NVL72 Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus GB200 NVL72 (NVIDIA Blackwell) on DeepSeekv4 Pro 0813 1.6T. Large-hyperscaler-volume ownership TCO normalized by total tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. 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 →
Push DeepSeekv4 Pro 0813 1.6T to 75 tok/s/user and B200 lands at $0.02 per million tokens against GB200 NVL72's $0.04 — B200 pulls ahead by 118%.
B200: $0.05 per million tokens. GB200 NVL72: $0.14. Both at 110 tok/s/user on DeepSeekv4 Pro 0813 1.6T, with B200 202% cheaper.
Toward the upper edge of the 41–180 tok/s/user interactivity band — at 145 tok/s/user — B200 runs $0.06 per million tokens on DeepSeekv4 Pro 0813 1.6T while GB200 NVL72 runs $0.19. B200 is the cheaper choice by 236%. (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.)
Chip pricing (owning hyperscaler): B200 $1.73/chip/hr · GB200 NVL72 $1.86/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.020GB200 NVL72:$0.043 | B200:$0.045GB200 NVL72:$0.136 | B200:$0.057GB200 NVL72:$0.192 |
| Concurrency | B200:~37GB200 NVL72:~38 | B200:~14GB200 NVL72:~5 | B200:~11GB200 NVL72:~3 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
DeepSeek V4 Pro 0813 1.6T 8K / 1K Cost per Million Total Tokens vs. Interactivity
DeepSeek V4 Pro 0813 1.6T 8K / 1K Cost per Million Total Tokens vs. Interactivity- Cost Tier:
- Owning at Large Hyperscaler Volume
- Source:
- SemiAnalysis InferenceX™
TCO $/chip/hr: VR200: 3.61H100: 1.17H200: 1.22B200: 1.73B300: 2.26GB200: 1.86GB300: 2.31MI300X: 0.95MI325X: 1.1MI355X: 1.5RTX6000PRO: 0.68Jalapeño (OpenAI): 1.47TPUV7: 1.03
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.
No data available
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