Kimi K3 2.8T · Chip comparison

Kimi K3 2.8T — GB300 NVL72 vs MI355X

Head-to-head AI inference benchmark comparison of GB300 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 →

GB300 NVL72 / MI355X on Kimi K3 2.8T at 79 tok/s/user: 17,040 / 7,014 tok/s/chip, $0.038 / $0.059 per million tokens. GB300 NVL72 is 37% cheaper per token; GB300 NVL72 delivers 143% more tok/s/chip.

Around the middle of the 35–211 tok/s/user interactivity band, at 123 tok/s/user on Kimi K3 2.8T: GB300 NVL72 runs 10,919 tok/s/chip at $0.059 per million tokens, MI355X runs 2,874 at $0.14. GB300 NVL72 is 59% cheaper per token; GB300 NVL72 delivers 3.8× the tok/s/chip.

Setting 167 tok/s/user as the target on Kimi K3 2.8T, GB300 NVL72 produces 9,625 tok/s/chip ($0.067 per million tokens) and MI355X produces 2,677 ($0.16). GB300 NVL72 is 57% cheaper per token; GB300 NVL72 delivers 3.6× the 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.)

View performance per dollar →

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)
GB300 NVL7217,040MI355X7,014
GB300 NVL7210,919MI355X2,874
GB300 NVL729,625MI355X2,677
Cost ($/M tok)
GB300 NVL72$0.038MI355X$0.059
GB300 NVL72$0.059MI355X$0.14
GB300 NVL72$0.067MI355X$0.16
tok/s/MW
GB300 NVL728,037,910MI355X3,356,129
GB300 NVL725,150,249MI355X1,375,139
GB300 NVL724,539,923MI355X1,280,724
Concurrency
GB300 NVL72~128MI355X~23
GB300 NVL72~41MI355X~4
GB300 NVL72~32MI355X~3