DeepSeek V3.2 API
About
Build advanced reasoning agents with DeepSeek V3.2, a specialized AI model optimized for complex logic and chain-of-thought tasks. This model delivers cost-effective pricing at $0.56/1M input and $1.68/1M output tokens, native tool calling support, open weights architecture. Access DeepSeek V3.2 via the Vertex API with up to 66K output tokens.
Capabilities
Input Modalities
textpdf
Output Modalities
text
Reasoning
Yes
Structured Output
Yes
Tool Use
Yes
WeightsOpen
Temperature
Adjustable
Attachment
Not Supported
Limits
Context Window
164K
Tokens
Input Limit
164K
Tokens
Max Output
66K
Tokens
Pricing
Standard (per 1M tokens)
Input
$0.56
Output
$1.68
Caching (per 1M tokens)
Read
$0.06
Frequently Asked Questions
DeepSeek V3.2 by Vertex costs $0.56 per 1M input tokens and $1.68 per 1M output tokens. Cached reads cost $0.06 per 1M tokens.
Model Details
ID
deepseek-ai/deepseek-v3.2-maas
Provider
VertexFamily
deepseek
Release Date
Dec 17, 2025
Knowledge Cutoff
N/A
API Integration
NPM Package
@ai-sdk/google-vertex
Environment Variables
GOOGLE_VERTEX_PROJECT
GOOGLE_VERTEX_LOCATION
GOOGLE_APPLICATION_CREDENTIALS