DeepSeek V4 Pro 降价：官方 token 价格与本站 Coding Plan 如何区分

中文摘要

DeepSeek V4 Pro 降价是当前价格主线。本站 Coding Plan 是一次性库存产品，token 官方价格与购买卡片价格需要分开阅读。

阅读提示

这篇中文稿保留原始来源链接，并把 DeepSeek 官方发布、报道和市场传闻分开标注。购买相关判断仍以 /zh/pricing 的真实库存卡片为准；出现在新闻或基准中的模型不代表可购买。

英文原文

Official pricing update

DeepSeek V4 Pro is no longer only a quality and long-context story. It is now the pricing headline. The current official pricing page shows the same low Pro rates and adds a stronger forward-looking note: after the 75% discount window ends at 2026-05-31 23:59 (Beijing time), V4-Pro API pricing will be officially adjusted to 1/4 of the original price.

What is officially confirmed

• DeepSeek V4 Pro cache-hit input: RMB 0.025 / 1M tokens
• DeepSeek V4 Pro cache-miss input: RMB 3 / 1M tokens
• DeepSeek V4 Pro output: RMB 6 / 1M tokens
• DeepSeek V4 Flash remains the high-volume route with lower non-Pro rates
• DeepSeek still reserves the right to adjust product pricing, so this should be described as an official quarter-price reset, not a forever-price promise

Why this matters

For teams with stable system prompts, repeated retrieval blocks, or heavy prompt-prefix reuse, the cache-hit line matters as much as the headline price cut. The practical result is that DeepSeek V4 Pro can stay in the serving path for harder coding and reasoning tasks without blowing up the budget.

Editorial takeaway

This is the new DeepSeek-first lead story for the homepage: lower V4 Pro pricing, much cheaper cache-hit traffic, and a stronger economic argument against defaulting to premium alternatives for every request.