Command Palette
Search for a command to run

DeepSeek V3.2

by Deepseek

Specifications

Input
Output
Context window
164K tokens
Released
Dec 2025

Performance

Speed
19 t/s
TTFT
3.3s
Latency
4.8s
Intelligence

Pricing

Input
€0.30
per 1M tokens
Output
€0.50
per 1M tokens
€0.08 cache hit

About this model

DeepSeek V3.2 is the latest iteration of the DeepSeek V3 series with significant performance improvements. Features enhanced reasoning, coding capabilities, and better instruction following across diverse tasks.

Technical specifications

Capabilities
Input modalities
Output modalities
Reasoning
Hybrid Default on

Knowledge horizon

Released Dec 2025
Today
Since release 8 mo

See also

Add Model to Comparison
Search for a model to add
Command Palette
Search for a command to run