Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on MiniMax M2.5/M2.7. 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.
Near the low end of the 24–178 tok/s/user interactivity band, at 62 tok/s/user on MiniMax M2.5/M2.7: B200 runs 12925 tok/s/chip at $0.04/M tokens, GB300 NVL72 runs 16544 at $0.04/M. B200 is 4% cheaper per token; GB300 NVL72 delivers 28% more tok/s/chip.
Setting 101 tok/s/user as the target on MiniMax M2.5/M2.7, B200 produces 6649 tok/s/chip ($0.07 per million tokens) and GB300 NVL72 produces 8079 ($0.08). B200 is 10% cheaper per token; GB300 NVL72 delivers 22% more tok/s/chip.
At 139 tok/s/user interactivity on MiniMax M2.5/M2.7, B200 delivers 3296 tok/s/chip at $0.15 per million tokens; GB300 NVL72 delivers 3804 tok/s/chip at $0.17. B200 is 16% cheaper per token; GB300 NVL72 delivers 15% more tok/s/chip at this point. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:12924.8GB300 NVL72:16544.4 | B200:6649.1GB300 NVL72:8078.6 | B200:3296.3GB300 NVL72:3804.3 |
| Cost ($/M tok) | B200:$0.037GB300 NVL72:$0.039 | B200:$0.072GB300 NVL72:$0.079 | B200:$0.146GB300 NVL72:$0.169 |
| tok/s/MW | B200:7558355GB300 NVL72:7803974 | B200:3888345GB300 NVL72:3810678 | B200:1927651GB300 NVL72:1794491 |
| Concurrency | B200:~876GB300 NVL72:~340 | B200:~9GB300 NVL72:~48 | B200:~15GB300 NVL72:~17 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
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 (Teacup): 1.47
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
