Kimi K2.5/K2.6/K2.7-Code 1T — B300 vs Jalapeño
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and Jalapeño (OpenAI Jalapeño) 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 / Jalapeño on Kimi K2.5/K2.6/K2.7-Code 1T at 89 tok/s/user: 2020 / 11518 tok/s/chip, $0.31 / $0.03 per million tokens. Jalapeño is 914% cheaper per token; Jalapeño delivers 470% more tok/s/chip.
Around the middle of the 61–173 tok/s/user interactivity band, at 117 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 runs 1196 tok/s/chip at $0.52/M tokens, Jalapeño runs 9997 at $0.04/M. Jalapeño is 1387% cheaper per token; Jalapeño delivers 736% more tok/s/chip.
Setting 145 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B300 produces 585 tok/s/chip ($1.07 per million tokens) and Jalapeño produces 8213 ($0.04). Jalapeño is 2398% cheaper per token; Jalapeño delivers 1303% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:2020.3Jalapeño:11517.6 | B300:1196.0Jalapeño:9997.1 | B300:585.2Jalapeño:8213.2 |
| Cost ($/M tok) | B300:$0.311Jalapeño:$0.031 | B300:$0.525Jalapeño:$0.035 | B300:$1.073Jalapeño:$0.043 |
| tok/s/MW | B300:1063329Jalapeño:10237897 | B300:629472Jalapeño:8886344 | B300:308005Jalapeño:7300618 |
| Concurrency | B300:~10Jalapeño:~1112 | B300:~5Jalapeño:~629 | B300:~2Jalapeño:~358 |