DeepSeek Launches V4.1 Flash — Back to $0.15/1M Input
$0.22/$0.44 input (off-peak/peak, V4 Flash)
$0.15/$0.30 input, $0.60/$1.20 output (V4.1 Flash, off-peak/peak)
What changed
Just 25 days after the peak/off-peak price increase, DeepSeek launched V4.1 Flash with dramatically lower rates. This is the real story: they raised prices, then cut them below the original flat rate within a month.
Before / After (Aug 16 → Sep 10)
| Aug 16 (V4 Flash) | Sep 10 (V4.1 Flash) | |
|---|---|---|
| Input off-peak (cache miss) | $0.22 | $0.15 |
| Input peak (cache miss) | $0.44 | $0.30 |
| Output off-peak | $0.66 | $0.60 |
| Output peak | $1.32 | $1.20 |
| Input (cache hit) | ~$0.004 | $0.003 |
Net result: V4.1 Flash at $0.15 input is actually cheaper than the original flat $0.14 for off-peak users, and competitive for peak users.
What this means for you
If you switched away after Aug 16: worth coming back. V4.1 Flash is genuinely cheap again.
If you're doing batch work: off-peak V4.1 Flash at $0.15/$0.60 is one of the cheapest frontier API options available.
What's next
On September 14, all deepseek-v4-pro requests will route to V4.1 Flash at V4.1 Flash rates. → Sep 14: V4 Pro routes to V4.1 Flash
Related
- DeepSeek pricing — current rates
- DeepSeek vs ChatGPT
Sources
Verified: September 12, 2026