Kimi K2.5/K2.6/K2.7-Code 1T — B300 vs H200
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) 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.
At 51 tok/s/user interactivity on Kimi K2.5/K2.6/K2.7-Code 1T, B300 delivers 464 tok/s/GPU at $1.43 per million tokens; H200 delivers 310 tok/s/GPU at $1.24. H200 is 16% cheaper per token; B300 delivers 49% more tok/s/GPU at this point.
B300 posts 304 tok/s/GPU for $2.13 per million tokens at 69 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T; H200 posts 229 tok/s/GPU for $1.70. H200 is 25% cheaper per token; B300 delivers 33% more tok/s/GPU.
Throughput at 88 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 hits 211 tok/s/GPU, H200 hits 158. Per-million costs land at $3.02 and $2.49 respectively. H200 is 21% cheaper per token; B300 delivers 33% more tok/s/GPU. (Numbers reflect the default 1k/1k · int4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:464.0H200:310.5 | B300:304.2H200:229.0 | B300:210.7H200:158.1 |
| Cost ($/M tok) | B300:$1.433H200:$1.238 | B300:$2.135H200:$1.704 | B300:$3.024H200:$2.493 |
| tok/s/MW | B300:213815H200:179479 | B300:140189H200:132358 | B300:97081H200:91376 |
| Concurrency | B300:~16H200:~25 | B300:~12H200:~13 | B300:~4H200:~7 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.