DeepSeek R1 Distill 32B

DeepSeek

Knowledge-distilled reasoning model from DeepSeek R1 (671B) into Qwen2.5-32B base. Chain-of-thought reasoning that outperforms OpenAI o1-mini. Most popular model on Ollama (79M+ pulls).

Text Generation Local DeepSeek Family vR1

Documentation Back to Models

Parameters

32B

params

Context Window

131K

tokens

Max Output

-

tokens

Input Price

-

per 1M tokens

Output Price

-

per 1M tokens

DeepSeek Family 15 models

The full DeepSeek line by generation — pricing and capabilities vary across the family.

V4

Agentic reasoning, coding, long-context

Cost-sensitive inference, coding, long-context

DeepSeek

General purpose, reasoning, code generation

Code generation, completion

Professional coding tasks

General purpose, MoE

General chat, coding

V3.2

Sparse attention, experimental

V3.1

Hybrid thinking, 40% better on SWE-bench

R1

Enhanced reasoning

R1 Distill 32B Current

Reasoning, math, coding, analysis

131K

context

Jan 2025

Reasoning, math, coding

Reasoning, chain-of-thought

V3

General purpose, MIT license

Coder

Code generation, debugging

Capabilities

👁️

Vision

⚡

Function Calling

📋

JSON Mode

🌊

Streaming

💬

System Prompt

🖥️

Code Execution

🔍

Web Search

🔌

MCP Support

Local Model Specs

Quantization

Q4_K_M

Architecture

Dense (Qwen2.5 base)

Runtime

Ollama / llama.cpp

Disk Size

20 GB

Details

Release Date: January 22, 2025
Knowledge Cutoff: -
Source: Local
License: Apache 2.0
Model ID: deepseek-r1-distill-32b

Last updated: March 13, 2026