Vipvamps/DeepSeek-R1-Distill-Qwen-1.5B

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 7, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

DeepSeek-R1-Distill-Qwen-1.5B is a 1.5 billion parameter language model developed by DeepSeek AI, distilled from the larger DeepSeek-R1 model and based on Qwen2.5-Math-1.5B. It is fine-tuned using reasoning data generated by DeepSeek-R1, excelling in mathematical, coding, and general reasoning tasks with a 32K context length. This model demonstrates that complex reasoning patterns can be effectively transferred to smaller, dense models, offering strong performance in a compact size.

Loading preview...

DeepSeek-R1-Distill-Qwen-1.5B: Distilled Reasoning Power

DeepSeek-R1-Distill-Qwen-1.5B is a 1.5 billion parameter model developed by DeepSeek AI, part of a series of distilled models derived from the larger DeepSeek-R1. DeepSeek-R1 itself is a first-generation reasoning model trained primarily through large-scale reinforcement learning (RL) without initial supervised fine-tuning (SFT), demonstrating advanced reasoning behaviors like self-verification and reflection.

Key Capabilities & Distillation Process

  • Reasoning Distillation: This model showcases DeepSeek AI's finding that complex reasoning patterns from larger models can be effectively distilled into smaller, dense models. It was fine-tuned using reasoning data generated by the powerful DeepSeek-R1.
  • Enhanced Performance: Despite its compact size, the model achieves strong performance across various benchmarks, particularly in mathematical and coding reasoning, outperforming many larger models in its class.
  • Qwen2.5 Base: Built upon the Qwen2.5-Math-1.5B architecture, it leverages the strengths of its base model while integrating advanced reasoning capabilities.
  • Context Length: Supports a substantial context length of 32,768 tokens, enabling it to handle complex and lengthy inputs.

When to Use This Model

  • Resource-Constrained Environments: Ideal for applications requiring strong reasoning capabilities where computational resources are limited, thanks to its efficient 1.5B parameter count.
  • Mathematical and Coding Tasks: Excels in benchmarks related to mathematics (e.g., AIME 2024, MATH-500) and coding (e.g., Codeforces, LiveCodeBench), making it suitable for these specialized domains.
  • Research and Development: Provides a valuable open-source option for researchers exploring model distillation techniques and the transfer of reasoning abilities to smaller LLMs.

Usage Recommendations

To achieve optimal performance, DeepSeek AI recommends specific configurations:

  • Set temperature between 0.5-0.7 (0.6 recommended).
  • Avoid system prompts; include all instructions in the user prompt.
  • For math problems, include "Please reason step by step, and put your final answer within \boxed{}" in the prompt.
  • Enforce the model to start its response with "\n" to ensure thorough reasoning.

Finetunes
1 models