Kimi K3 2.8T · Chip comparison
Kimi K3 2.8T — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Kimi K3 2.8T. 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 →
At 71 tok/s/user interactivity on Kimi K3 2.8T, GB200 NVL72 delivers 7,614 tok/s/chip at $0.068 per million tokens; MI355X delivers 8,284 tok/s/chip at $0.050. MI355X is 26% cheaper per token; MI355X delivers 9% more tok/s/chip at this point.
GB200 NVL72 posts 3,135 tok/s/chip for $0.16 per million tokens at 118 tok/s/user on Kimi K3 2.8T; MI355X posts 3,024 tok/s/chip for $0.14. MI355X is 16% cheaper per token; GB200 NVL72 delivers 4% more tok/s/chip.
Throughput at 165 tok/s/user on Kimi K3 2.8T: GB200 NVL72 hits 2,298 tok/s/chip, MI355X hits 2,700. Costs per million tokens land at $0.22 and $0.15 respectively. MI355X is 31% cheaper per token; MI355X delivers 17% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL727,614MI355X8,284 | GB200 NVL723,135MI355X3,024 | GB200 NVL722,298MI355X2,700 |
| Cost ($/M tok) | GB200 NVL72$0.068MI355X$0.050 | GB200 NVL72$0.16MI355X$0.14 | GB200 NVL72$0.22MI355X$0.15 |
| tok/s/MW | GB200 NVL724,071,420MI355X3,963,786 | GB200 NVL721,676,393MI355X1,447,072 | GB200 NVL721,229,039MI355X1,291,953 |
| Concurrency | GB200 NVL72~38MI355X~44 | GB200 NVL72~8MI355X~7 | GB200 NVL72~8MI355X~4 |