NEW · API PLANS - 9 models, one key, 250M - 3B tokens a month · from /mo See plans →
Home › Models › DeepSeek › DeepSeek Flash

DeepSeek Flash

DeepSeek Flash is the high-volume endpoint in the DeepSeek family. It is a 671B model with a 1M context window and accepts text and image input. Use it for classification, field extraction, and short edits where token cost matters more than reasoning depth. Its intelligence index here is 19, the lowest DeepSeek score in the catalog, so it is the wrong route for algorithms, multi-file refactors, or anything that needs a careful chain of thought.

ClassificationExtractionHigh volumeLow latencyHigh-Volume Utility Tasks

Live throughput history

source: status.ebotservers.com
Parameters671B
Context1M
Intelligence index19
TPS avg 24h0.0
TTFB p500 ms
Uptime100.0%

Call it


 
 

Other models from DeepSeek