Kimi K2.5/K2.6/K2.7-Code 1T — B300 vs MI355X
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Kimi K2.5/K2.6/K2.7-Code 1T. 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.
B300 / MI355X on Kimi K2.5/K2.6/K2.7-Code 1T at 54 tok/s/user: 1059 / 632 tok/s/GPU, $0.61 / $0.66 per million tokens. B300 is 8% cheaper per token; B300 delivers 68% more tok/s/GPU.
Around the middle of the 35–113 tok/s/user interactivity band, at 74 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 runs 925 tok/s/GPU at $0.70/M tokens, MI355X runs 516 at $0.80/M. B300 is 13% cheaper per token; B300 delivers 79% more tok/s/GPU.
Setting 94 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B300 produces 910 tok/s/GPU ($0.72 per million tokens) and MI355X produces 396 ($1.04). B300 is 45% cheaper per token; B300 delivers 130% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:1059.3MI355X:631.8 | B300:925.0MI355X:516.1 | B300:910.0MI355X:396.0 |
| Cost ($/M tok) | B300:$0.612MI355X:$0.658 | B300:$0.704MI355X:$0.797 | B300:$0.715MI355X:$1.040 |
| tok/s/MW | B300:488139MI355X:238416 | B300:426266MI355X:194767 | B300:419364MI355X:149436 |
| Concurrency | B300:~40MI355X:~24 | B300:~26MI355X:~14 | B300:~21MI355X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.