DeepSeek slashes Flash pricing up to 57%, retires V4 Pro Sept 14
What happened
Between the Sep 10, 2026 and Sep 11, 2026 captures of DeepSeek's pricing page, Pulse recorded 2 changes: Flash tier token prices cut across the board (cache-hit input -57%, cache-miss input -32%, output -9%) while the separate flash-vision-exp model was retired and merged into the base Flash model, which now natively supports Vision. Pro tier pricing unchanged; New notice: DeepSeek will retire the V4 Pro model starting Sept 14, 2026 12:00 Beijing Time, rerouting all deepseek-v4-pro requests to V4.1. In detail: New notice: DeepSeek will retire the V4 Pro model starting Sept 14, 2026 12:00 Beijing Time, rerouting all deepseek-v4-pro requests to V4.1 Flash and billing them at the (cheaper) Flash price until a V4.1 Pro model ships. Pulse files it as a pricing change (price decrease, packaging change) with a downward direction. Pulse's latest capture (Nov 15, 2026) lists DeepSeek Flash at $0.15 per 1M input tokens (cache miss) + $0.30 per 100M input tokens (cache hit) and DeepSeek V4 Pro at $0.66 per 1M input tokens (cache miss) + $2.20 per 100M input tokens (cache hit).
It lands 18 days after DeepSeek's previous change, "DeepSeek adds vision model, exempts weekends from peak pricing" (Aug 23, 2026). Before that Pulse recorded "DeepSeek: API pricing restructured to peak/off-peak billing, prices roughly double" (Aug 17, 2026), "DeepSeek: dropped peak/off-peak pricing plan, warns of broader increase" (Aug 10, 2026) and "DeepSeek: announced peak/off-peak surge pricing (2x during peak hours)" (Aug 3, 2026). DeepSeek has been tracked since 2026-05-16.
In the same window OpenAI, Google, X, Metabase and Mistral AI also changed their AI & Machine Learning pricing pages. Those moves are listed further down, and the Pulse Agent can compare them side by side.
- RemovedFlashPriceFlash tier token pricesNot listed
- Packaging changeFlashPackaging
New notice: DeepSeek will retire the V4 Pro model starting Sept 14, 2026 12:00 Beijing Time, rerouting all deepseek-v4-pro requests to V4.1 Flash and billing them at the (cheaper) Flash price until a V4.1 Pro model ships.
| Plan | Field | Before | After | Change |
|---|---|---|---|---|
| Flash | Price | Flash tier token prices | Not listed | Removed |
| Flash | Packaging | New notice: DeepSeek will retire the V4 Pro model starting Sept 14, 2026 12:00 Beijing Time, rerouting all deepseek-v4-pro requests to V4.1 Flash and billing them at the (cheaper) Flash price until a V4.1 Pro model ships. | Packaging change | |
Values as written on DeepSeek's public pricing page in the two captures Pulse compared; where the page states no number, the cells give the wording before and after.
Evidence
The two captures Pulse comparedFull-page captures of DeepSeek's pricing page (deepseek.com) from Sep 10, 2026 and Sep 11, 2026; the change is marked around the middle of the page, where the Flash tier token prices no longer appears.
Sign in to see the evidence
The two full-page captures with this change marked up are free with an account.
DeepSeek, the last 90 days
Full history- Sep 20, 2026PackagingDeepSeek extends 50% off-peak API discount to Chinese public holidays all day
- Sep 13, 2026PackagingDeepSeek merges v4-flash and vision-exp into cheaper Flash V4.1
- Sep 12, 2026PackagingDeepSeek reverses V4 Pro retirement, keeps API billing unchanged past Sept 14
- Sep 11, 2026Price decreaseDeepSeek slashes Flash pricing up to 57%, retires V4 Pro Sept 14 This change
- Aug 23, 2026PackagingDeepSeek adds vision model, exempts weekends from peak pricing
- Aug 17, 2026Price increaseDeepSeek: API pricing restructured to peak/off-peak billing, prices roughly double
- Aug 10, 2026PackagingDeepSeek: dropped peak/off-peak pricing plan, warns of broader increase
- Aug 3, 2026PackagingDeepSeek: announced peak/off-peak surge pricing (2x during peak hours)
Across AI & Machine Learning in the same window
All AI & Machine Learning changes- OpenAISep 11OpenAI quietly drops 20x usage option from Pro, rewrites plan taglines
- GoogleSep 7Google Workspace: promo discount cut from 50% to 30%/20% off
- XAug 24X drops grok-4.5, adds Imagine image/video models and free Playground perk
- MetabaseAug 24Metabase quietly reveals Enterprise-only paid air-gapped deployment option
- Mistral AIAug 17Mistral AI folds Education into Pro, adds bundled API credits to entry plans
Questions about this change
Did DeepSeek raise prices in September 2026?
No — prices went down. Between Sep 10, 2026 and Sep 11, 2026 DeepSeek's pricing page showed 1 price cut: Flash: Flash tier token prices → Not listed (removed).
What changed on DeepSeek's pricing page between Sep 10, 2026 and Sep 11, 2026?
Flash price removed: Flash tier token prices; New notice: DeepSeek will retire the V4 Pro model starting Sept 14, 2026 12:00 Beijing Time, rerouting all deepseek-v4-pro requests to V4.1 Flash and billing them at the (cheaper) Flash price until a V4.1 Pro model ships. Pulse compared the captures it took on Sep 10, 2026 and Sep 11, 2026; the before/after table above lists each item as written on the page.
How often does DeepSeek change its pricing?
Pulse recorded 4 other pricing-page changes for DeepSeek in the 90 days before this one; the previous change was "DeepSeek adds vision model, exempts weekends from peak pricing" on Aug 23, 2026; DeepSeek has been tracked since 2026-05-16.