← DeepSeek family

DeepSeek-R1-Distill-Qwen-7B

A small reasoning model distilled from DeepSeek-R1 onto Qwen2.5.

Overview

DeepSeek-R1-Distill-Qwen-7B is a small reasoning model made by fine-tuning Qwen2.5-Math-7B on answers from DeepSeek-R1. It thinks step by step before answering and is especially good at maths, in a model small enough to run on a single GPU or a CPU. It's released under the MIT licence.

Model specifications
WeightsOpen
Parameters7.6B
Context window128K tokens
LicenceMIT
Commercial useYes

LLM hosts that offer DeepSeek-R1-Distill-Qwen-7B

All hosts →
LLM hostHosting planHosting typePricing
xCloudxCloud LocalAI HostingGPU cloudFrom US$59.99Monthly subscription
xCloudxCloud Ollama HostingGPU cloudFrom US$59.99Monthly subscription