christophersmith2737-commits/OffPeak

README 摘要

OffPeak · dsh-offpeak Peak-hour price guard for DeepSeek API tidal pricing Intercept your message before it's sent during peak hours — schedule it to run when it's cheaper. Read this in 中文 Why DeepSeek moved to peak/off-peak (tidal) API pricing effective 2026-08-17 (Beijing time): Period Hours (Beijing) Output price vs. before Peak 09:00–12:00, 14:00–18:00 up to 4.5× Off-peak everything else half of peak New per-million-token rates (¥): Model Item Peak Off-peak V4 Flash input (cache miss) 3 1.5 output 9 4.5 input (cache hit) 0.1 0.05 V4 Pro input (cache miss) 9 4.5 output 27 13.5 input (cache hit) 0.3 0.15 If you mostly run long tasks with DeepSeek V4 Flash, sending them at 10:00 costs 3× more than sending them at 19:00. OffPeak makes sure you notice that before you press send — not after. What it does While your current model is DeepSeek V4 Flash / Pro and the Beijing clock is inside a peak window, pressing Enter (or clicking send) in the composer intercepts the message before it is sent . The text stays in the input box and a dialog appears: - Current price card — peak vs. off-peak per-million-token rates for the model in use - Continue — sends the message now, through the normal…

分类