DeepSeek V3.2 Exp

ReasoningTool CallingOpen WeightsStructured Output

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

Providers10

ReleasedJan 1, 2024

Input Modalitiestext

Output Modalitiestext

Tarsk Usecoding

Benchmarks

Intelligence Index

Math Index

Available Providers (10)

Provider	Model ID	Input Cost	Output Cost	Context	Max Output
iFlow	`deepseek-v3.2`	$0/MTok	$0/MTok	128K	64K
ZenMux	`deepseek/deepseek-v3.2-exp`	$0.22/MTok	$0.33/MTok	163K	64K
NovitaAI	`deepseek/deepseek-v3.2-exp`	$0.27/MTok	$0.41/MTok	163.8K	65.5K
Kilo Gateway	`deepseek/deepseek-v3.2-exp`	$0.27/MTok	$0.41/MTok	163.8K	65.5K
OpenRouter	`deepseek/deepseek-v3.2-exp`	$0.27/MTok	$0.41/MTok	163.8K	65.5K
SiliconFlow	`deepseek-ai/DeepSeek-V3.2-Exp`	$0.27/MTok	$0.41/MTok	164K	164K
Meganova	`deepseek-ai/DeepSeek-V3.2-Exp`	$0.27/MTok	$0.40/MTok	164K	164K
NanoGPT	`deepseek-ai/deepseek-v3.2-exp`	$0.28/MTok	$0.42/MTok	163.8K	65.5K
Alibaba (China)	`deepseek-v3-2-exp`	$0.29/MTok	$0.43/MTok	131.1K	65.5K
Qiniu	`deepseek/deepseek-v3.2-exp`	—/MTok	—/MTok	128K	32K

Capabilities

Reasoning

Tool Calling

Attachments

Open Weights

Structured Output