OPEN_SOURCE ↗
REDDIT · REDDIT// 3h agoPRODUCT UPDATE
DeepSeek slashes V4-Pro prices 75%
DeepSeek is offering a limited-time 75% discount on V4-Pro and cutting input cache-hit pricing across its API to one-tenth of launch levels. The promo runs through May 31, 2026, and makes an already aggressive frontier-model price even harder to ignore.
// ANALYSIS
This is classic DeepSeek pricing warfare: instead of competing only on benchmark bragging rights, it is using margin pressure as a product feature.
- –V4-Pro is already DeepSeek’s premium model, so a 75% promo lowers the trial barrier for teams that were curious but cost-sensitive
- –The cache-hit reduction matters for agentic, RAG, and long-context workloads where repeated prefixes can dominate spend
- –The move pressures rival API providers to defend premium pricing with quality, latency, or tooling advantages rather than raw model cost
- –The limited-time framing suggests user acquisition and workload capture, not a permanent reset of the price floor
// TAGS
deepseekdeepseek-v4-propricingllmapiinference
DISCOVERED
3h ago
2026-04-28
PUBLISHED
6h ago
2026-04-28
RELEVANCE
9/ 10
AUTHOR
elemental-mind