TL;DR
DeepSeek officially released V4.1 Flash on September 10, 2026, 12:00 Beijing time, with new Flash-series API pricing: cached input dropped 60% to 0.02 CNY per million tokens and output is 4 CNY per million tokens; V4 Pro retires on September 14.
Event overview
On September 10, 2026, DeepSeek notified API users that it was officially releasing V4.1 Flash, with the new Flash-series pricing effective at 12:00 Beijing time the same day. The company said that after internal and external testing, V4.1 Flash outperforms the previous V4 Pro on performance, cost, speed and total time. DeepSeek had announced the price adjustment on its official platform on September 9, 2026. The price change applies only to the Flash series; the higher-tier V4 Pro keeps its price and will be retired on September 14, 2026.
Key facts
| Item | Detail |
|---|---|
| Announcement | DeepSeek open platform, September 9, 2026 |
| Effective time | September 10, 2026, 12:00 Beijing time |
| Input (cache hit) | 0.05 → 0.02 CNY per million tokens, down 60% |
| Input (cache miss) | 1.5 → 1 CNY per million tokens, down about 33% |
| Output | 4 CNY per million tokens (off-peak) |
| Peak hours | Mon–Fri 9:00–12:00 and 14:00–18:00 Beijing time, priced at 2x off-peak |
| V4 Pro | Price unchanged; retired September 14, 2026 |