Summary
DeepSeek updated its API pricing on April 26, 2026 so cache-hit input pricing across all models falls to one-tenth of launch pricing. The same pricing page also shows 1M-context V4 Flash and V4 Pro pricing, reinforcing DeepSeek's strategy of competing on long-context cost efficiency as much as raw model quality.
What changed
DeepSeek reduced input cache-hit pricing to one-tenth of launch price across its API lineup, effective April 26, 2026.
Why it matters
Long-context and repeated-prefix workloads are a major cost center for agent and coding products. A permanent cache-hit reduction changes the economics of replay-heavy agent loops, retrieval-heavy prompts, and enterprise copilots that repeatedly ship large system context on every turn.
Evidence excerpt
DeepSeek's pricing page says that for all models, the input cache hit price has been reduced to 1/10 of the launch price effective from 2026/4/26 12:15 UTC.