DeepSeek V4-Flash-Vision-Exp
Last verified September 10, 2026 · DeepSeek pricing ↗$0.220/1M input · $0.660/1M output · 1.0M context · DeepSeek
Retired
Not generally available
DeepSeek has shut DeepSeek V4-Flash-Vision-Exp down. You cannot buy it at the rate below. The price is kept so past spend and old estimates still reconcile, and it is excluded from the calculator’s default model list and from every ranking.
DeepSeek’s own page ↗Count Tokens
—
Tokens
—
Input Cost
—
Output Cost
Estimate Monthly Cost
Monthly Cost Estimator
Quick:
<$0.0001/mo
Pick a preset above or enter custom usage
Alternatives to DeepSeek V4-Flash-Vision-Exp
Pricing Details
DeepSeekRETIRED 2026-09-10 after twenty days. The name still resolves but is 'temporarily routed' to DeepSeek V4.1 Flash (model name deepseek-flash), which has native vision and bills $0.15 / $0.003 / $0.60 off-peak. The card below is the last one this model was sold at. The image-token figures in this note were measured against the V4-Flash vision calculator and may not describe V4.1 Flash's tokenizer. DeepSeek's first vision model on the metered API, released 2026-08-21 and labelled experimental. Bills on exactly the deepseek-v4-flash card with no multimodal surcharge: off-peak $0.22 input / $0.66 output / $0.007 cache hit, peak $0.44 / $1.32 / $0.014. As with the text model the stored figure is the OFF-PEAK rate, which applies 17 of 24 hours; peak is 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday only, so off-peak is the majority case rather than the exception. Images are metered as input tokens with a hard ceiling: DeepSeek's docs state "Images are tokenized for billing: up to 384 tokens each, at V4-Flash pricing". Every image is resized before inference, up to roughly the pixel count of an 800x800 image, so a 2000x2000 and a 5000x5000 image cost the same. The 384 figure is a ceiling and not a flat charge, and a 16:9 screenshot does not reach it: the calculator DeepSeek ships on its own docs site returns 369 tokens for 1920x1080, 2560x1440 and 3840x2160 alike, and 349 for any square image at or above 1000x1000. DeepSeek calls that calculator an estimate and names the API usage field as the source of truth. A detail parameter exists: low downscales to 512x512, while high, original and auto all keep the original image (auto currently equals original). Limits are 8192 px per side, dropping to 4096 px once a request carries 15 or more images, 600 images per request, and 32 MiB per inline image. Cheapest per image of any vision model we have priced: at 369 tokens and the off-peak rate, one thousand 4K screenshots cost $0.0812.
Input / 1M tokens
$0.22
Output / 1M tokens
$0.66
Context Window
1.0M
Max Output
384K
Price History
No price change recorded since 2026-08-26, when this site started tracking this model; it launched on 2026-08-21.
Frequently Asked Questions
Common questions about DeepSeek V4-Flash-Vision-Exp pricing and usage