Kimi K3 2.8T — H200 vs MI355X Performance per Dollar
Cost per million tokens of H200 (NVIDIA Hopper) versus MI355X (AMD CDNA 4) on Kimi K3 2.8T. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. 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 →
MI355X edges H200 at 29 tok/s/user on Kimi K3 2.8T — $0.09 per million tokens versus $4.85, a 5202% cost-per-token gap.
Push Kimi K3 2.8T to 57 tok/s/user and H200 lands at $4.85 per million tokens against MI355X's $0.14 — MI355X pulls ahead by 3281%.
H200: $4.85 per million tokens. MI355X: $0.21. Both at 84 tok/s/user on Kimi K3 2.8T, with MI355X 2180% cheaper. (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.)
Chip pricing (owning hyperscaler): H200 $1.22/chip/hr · MI355X $1.50/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | H200:$4.850MI355X:$0.091 | H200:$4.850MI355X:$0.143 | H200:$4.850MI355X:$0.213 |
| Concurrency | H200:~1MI355X:~10 | H200:~1MI355X:~6 | H200:~1MI355X:~3 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.