DeepSeek adds Responses API to V4-Flash, announces future peak-hour surcharge

DeepSeek’s official Models & Pricing page (api-docs.deepseek.com) picked up two structural additions between the 2026-07-29 and 2026-08-05 captures, while every per-token rate stayed unchanged ($0.14/$0.0028/$0.28 for V4-Flash; $0.435/$0.003625/$0.87 for V4-Pro, cache-miss/cache-hit/output per 1M tokens).

First, a Responses API feature row now appears on the model comparison table, live today for deepseek-v4-flash only — the docs note deepseek-v4-pro support is coming “in early August 2026.” Second, and more consequential, DeepSeek disclosed a forthcoming time-of-use pricing policy: during peak hours (9:00-12:00 and 14:00-18:00 Beijing Time daily), “prices will be 2x the regular prices, applicable to all billing items.” No effective date has been set — DeepSeek says it is “subject to the official announcement” — so today’s flat rate card still applies. This reverses the direction of DeepSeek’s now-discontinued 2025-era off-peak discount into a peak-hour surcharge, a notable shift for a company whose pricing has so far moved in only one direction: down.