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 58 tok/s/user: 973 / 930 tok/s/GPU, $0.67 / $0.44 per million tokens. MI355X is 51% cheaper per token; B300 delivers 5% more tok/s/GPU.
Around the middle of the 35–128 tok/s/user interactivity band, at 81 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 runs 919 tok/s/GPU at $0.71/M tokens, MI355X runs 648 at $0.63/M. MI355X is 12% cheaper per token; B300 delivers 42% more tok/s/GPU.
Setting 105 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B300 produces 898 tok/s/GPU ($0.72 per million tokens) and MI355X produces 460 ($0.88). B300 is 21% cheaper per token; B300 delivers 95% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:972.7MI355X:930.1 | B300:918.7MI355X:647.8 | B300:898.0MI355X:460.2 |
| Cost ($/M tok) | B300:$0.668MI355X:$0.443 | B300:$0.708MI355X:$0.635 | B300:$0.725MI355X:$0.878 |
| tok/s/MW | B300:511922MI355X:445048 | B300:483549MI355X:309938 | B300:472625MI355X:220168 |
| Concurrency | B300:~34MI355X:~33 | B300:~24MI355X:~16 | B300:~18MI355X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.