DeepSeek slashes Flash pricing up to 57%, retires V4 Pro Sept 14
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
What changed
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 Flash and billing them at the (cheaper) Flash price until a V4.1 Pro model ships.
Before and after
Sep 10, 2026 vs Sep 11, 2026Flash 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.
Sign in to view the full comparison
See the before/after screenshots and every marked-up change.
More from DeepSeek
Full history →- Week 39, 2026·packagingDeepSeek: peak-hour billing now excludes Chinese public holidays
- Week 38, 2026·packagingDeepSeek merges v4-flash and vision-exp into cheaper Flash V4.1
- Week 35, 2026·packagingDeepSeek adds vision model, exempts weekends from peak pricing
- Week 34, 2026·increaseDeepSeek: API pricing restructured to peak/off-peak billing, prices roughly double
- Week 33, 2026·packagingDeepSeek: dropped peak/off-peak pricing plan, warns of broader increase