DeepSeek releases V4-vision-exp with extreme speed and low cost

teortaxesTex · x · 2026-08-21

DeepSeek released V4-vision-exp, consuming 117-384 tokens per image at V4-Flash prices. While it may not be the top VLM capability-wise, it offers magical speed and cost efficiency, posing a challenge to inference fleets.

Related event: DeepSeek Launches V4-vision-exp: Fast and Low-Cost Vision Model(2 posts)→

Original post →

More from Models

Models channel →