DeepSeek-V3.2

Caution: As of July 21, 2026, the deepseek-v3.2-maas endpoint is deprecated and will be retired on October 21, 2026. For more information, see Open model deprecations.

DeepSeek-V3.2 is a model that harmonizes high computational efficiency with superior reasoning and agent performance. DeepSeek's approach is built upon three key technical breakthroughs: DeepSeek Sparse Attention (DSA), scalable reinforcement learning framework, and large scale agentic task synthesis pipeline.

Managed API (MaaS) specifications

View pricing

Model ID	deepseek-v3.2-maas
Modalities	description Text Input and output photo Image Not supported mic Audio Not supported videocam Video Not supported
Capabilities	Function calling Supported Structured output Supported Thinking Not supported
Consumption options	Provisioned Throughput Supported Batch inference Supported Pay-as-you-go Standard PayGo Supported Fixed quota Not supported
Supported regions	Model availability	Global: global
	ML processing	Multi-region: us
Quotas	global: 65,536 maximum output, 163,840 context length
Versions	deepseek-v3.2-maas Launch stage: GA Release date: December 10, 2025

Deploy as a self-deployed model

To self-deploy the model, navigate to the DeepSeek-V3.2 model card in the Model Garden console and click Deploy model. For more information about deploying and using partner models, see Deploy a partner model and make prediction requests.