Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and H100 (NVIDIA Hopper) on gpt-oss 120B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β same interactions as the main inference chart.
Throughput at 90 tok/s/user on gpt-oss 120B: B200 hits 36090 tok/s/chip, H100 hits 6362. Per-million costs land at $0.01 and $0.05 respectively. B200 is 284% cheaper per token; B200 delivers 467% more tok/s/chip.
B200 / H100 on gpt-oss 120B at 143 tok/s/user: 24876 / 4148 tok/s/chip, $0.02 / $0.08 per million tokens. B200 is 306% cheaper per token; B200 delivers 500% more tok/s/chip.
Toward the upper edge of the 37β250 tok/s/user interactivity band, at 197 tok/s/user on gpt-oss 120B: B200 runs 17465 tok/s/chip at $0.03/M tokens, H100 runs 2636 at $0.12/M. B200 is 348% cheaper per token; B200 delivers 563% more tok/s/chip. (Numbers reflect the default 8k/1k Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:36089.7H100:6361.9 | B200:24875.7H100:4147.7 | B200:17464.7H100:2636.0 |
| Cost ($/M tok) | B200:$0.013H100:$0.051 | B200:$0.019H100:$0.078 | B200:$0.028H100:$0.123 |
| tok/s/MW | B200:21105097H100:4643748 | B200:14547213H100:3027501 | B200:10213266H100:1924056 |
| Concurrency | B200:~42H100:~32 | B200:~32H100:~13 | B200:~17H100:~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 β’ Drag to pan β’ Double-click to reset β’ Click a point to pin tooltip
gpt-oss 120B 8K / 1K Total Tokens per $1 TCO vs. Interactivity