DeepSeek V4 Flash 0731 reasoning

openrouter
deepseek
DeepSeek V4 Flash 0731 is a sparse mixture-of-experts model from DeepSeek, with 13B active parameters out of 284B total. This re-post-trained revision is suited for coding, reasoning, and agent workflows.

Capabilities

Context Window 1M tokens
Max Output 384k tokens
Inputs
Outputs

Pricing (per 1M tokens)

Input $0.14
Output $0.28
Cache Read $0.00
Cache Write -

Supported Parameters

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Move Chat

Select a destination folder:

Move Folder

Select a parent folder:

Folder Settings