Kimi K3.1 Pro — GB300 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Kimi K3.1 Pro. 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 K3.1 Pro at 90 tok/s/user: 20223 / 5179 tok/s/chip, $0.03 / $0.08 per million tokens. GB300 NVL72 is 154% cheaper per token; GB300 NVL72 delivers 290% more tok/s/chip.
Around the middle of the 62–175 tok/s/user interactivity band, at 118 tok/s/user on Kimi K3.1 Pro: GB300 NVL72 runs 13258 tok/s/chip at $0.05/M tokens, MI355X runs 3400 at $0.12/M. GB300 NVL72 is 153% cheaper per token; GB300 NVL72 delivers 290% more tok/s/chip.
Setting 147 tok/s/user as the target on Kimi K3.1 Pro, GB300 NVL72 produces 9632 tok/s/chip ($0.07 per million tokens) and MI355X produces 2539 ($0.16). GB300 NVL72 is 146% cheaper per token; GB300 NVL72 delivers 279% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:20223.1MI355X:5179.0 | GB300 NVL72:13258.4MI355X:3399.7 | GB300 NVL72:9632.0MI355X:2538.6 |
| Cost ($/M tok) | GB300 NVL72:$0.032MI355X:$0.080 | GB300 NVL72:$0.048MI355X:$0.123 | GB300 NVL72:$0.067MI355X:$0.164 |
| tok/s/MW | GB300 NVL72:9539197MI355X:2477998 | GB300 NVL72:6253975MI355X:1626674 | GB300 NVL72:4543402MI355X:1214636 |
| Concurrency | GB300 NVL72:~101MI355X:~8 | GB300 NVL72:~64MI355X:~5 | GB300 NVL72:~47MI355X:~3 |