Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and H200 (NVIDIA Hopper) on Qwen 3.8 Flash Next 176B-A6B. 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.
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 β
Setting 46 tok/s/user as the target on Qwen 3.8 Flash Next 176B-A6B, H100 produces 4980 tok/s/chip ($0.07 per million tokens) and H200 produces 16937 ($0.02). H200 is 226% cheaper per token; H200 delivers 240% more tok/s/chip.
At 77 tok/s/user interactivity on Qwen 3.8 Flash Next 176B-A6B, H100 delivers 4944 tok/s/chip at $0.07 per million tokens; H200 delivers 15499 tok/s/chip at $0.02. H200 is 201% cheaper per token; H200 delivers 213% more tok/s/chip at this point.
H100 posts 4908 tok/s/chip for $0.07 per million tokens at 108 tok/s/user on Qwen 3.8 Flash Next 176B-A6B; H200 posts 13006 tok/s/chip for $0.03. H200 is 154% cheaper per token; H200 delivers 165% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp8 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H100:4979.7H200:16937.4 | H100:4944.2H200:15499.1 | H100:4907.6H200:13006.3 |
| Cost ($/M tok) | H100:$0.065H200:$0.020 | H100:$0.066H200:$0.022 | H100:$0.066H200:$0.026 |
| tok/s/MW | H100:3634841H200:12363057 | H100:3608888H200:11313215 | H100:3582201H200:9493660 |
| Concurrency | H100:~11H200:~15 | H100:~10H200:~13 | H100:~9H200:~11 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
