DeepSeek V3.2 by Nvidia — Specs, Pricing & Capabilities | whichllm

Model ID deepseek-ai/deepseek-v3.2

Provider Nvidia

Family deepseek

Status -

Knowledge Cutoff 2024-07

Release Date 2025-12-01

Input Modalities text

Output Modalities text

Context Window 163840

Input Limit -

Output Limit 65536

Tool Calling Yes

Reasoning Yes

Structured Output -

Temperature Control Yes

Open Weights No

Input Cost / 1M tokens -

Output Cost / 1M tokens -

Reasoning Cost / 1M tokens -

Cache Read Cost / 1M tokens -

Cache Write Cost / 1M tokens -