Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI355X (AMD CDNA 4) on DeepSeek V4.1 Flash 552B. 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.
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 →
Near the low end of the 7–303 tok/s/user interactivity band, at 81 tok/s/user on DeepSeek V4.1 Flash 552B: H100 runs 10947 tok/s/chip at $0.03/M tokens, MI355X runs 83579 at $0.00/M. MI355X is 496% cheaper per token; MI355X delivers 664% more tok/s/chip.
Setting 155 tok/s/user as the target on DeepSeek V4.1 Flash 552B, H100 produces 6922 tok/s/chip ($0.05 per million tokens) and MI355X produces 45735 ($0.01). MI355X is 415% cheaper per token; MI355X delivers 561% more tok/s/chip.
At 230 tok/s/user interactivity on DeepSeek V4.1 Flash 552B, H100 delivers 4187 tok/s/chip at $0.08 per million tokens; MI355X delivers 33074 tok/s/chip at $0.01. MI355X is 516% cheaper per token; MI355X delivers 690% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H100:10946.7MI355X:83578.6 | H100:6921.5MI355X:45735.0 | H100:4187.3MI355X:33073.8 |
| Cost ($/M tok) | H100:$0.030MI355X:$0.005 | H100:$0.047MI355X:$0.009 | H100:$0.078MI355X:$0.013 |
| tok/s/MW | H100:7990287MI355X:39989751 | H100:5052199MI355X:21882753 | H100:3056440MI355X:15824787 |
| Concurrency | H100:~15MI355X:~26 | H100:~8MI355X:~13 | H100:~4MI355X:~9 |
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.
