Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) on GLM 5/5.1. 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 18–84 tok/s/user interactivity band, at 34 tok/s/user on GLM 5/5.1: B300 runs 1310 tok/s/chip at $0.48/M tokens, H200 runs 740 at $0.46/M. H200 is 5% cheaper per token; B300 delivers 77% more tok/s/chip.
Setting 51 tok/s/user as the target on GLM 5/5.1, B300 produces 1085 tok/s/chip ($0.58 per million tokens) and H200 produces 563 ($0.60). B300 is 4% cheaper per token; B300 delivers 93% more tok/s/chip.
At 68 tok/s/user interactivity on GLM 5/5.1, B300 delivers 885 tok/s/chip at $0.71 per million tokens; H200 delivers 434 tok/s/chip at $0.78. B300 is 10% cheaper per token; B300 delivers 104% more tok/s/chip at this point. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:1310.5H200:740.0 | B300:1084.5H200:563.3 | B300:885.1H200:433.9 |
| Cost ($/M tok) | B300:$0.479H200:$0.458 | B300:$0.579H200:$0.602 | B300:$0.709H200:$0.781 |
| tok/s/MW | B300:689711H200:540170 | B300:570799H200:411175 | B300:465841H200:316738 |
| Concurrency | B300:~36H200:~21 | B300:~20H200:~10 | B300:~12H200:~6 |
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
GLM5/5.1 744B 8K / 1K Total Tokens per $1 TCO vs. Interactivity