DeepSeek-R1-Distill-Qwen-14B

by DeepSeek · DeepSeek R1 family · released Jan 20, 2025
Open weights Active

14B distill.

This model has no first-party API. Open-weight release: self-host from published weights.

Specifications

Model typeText LLM
ArchitectureDense
Parameters14
Context window131.1K tokens
Max output8.2K tokens
Knowledge cutoffAug 1, 2024
LicenseMIT License
Modalities Text input Text output
Tags Distilled Reasoning Small LM
End of lifeNot disclosed
Last verifiedAug 20, 2026