NostraEmpire/mirror-deepseek-r1-distill-qwen-7b

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 31, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

NostraEmpire/mirror-deepseek-r1-distill-qwen-7b is a 7.6 billion parameter language model from DeepSeek-AI, distilled from the larger DeepSeek-R1 model and based on Qwen2.5-Math-7B. It is specifically fine-tuned to excel in reasoning tasks across math, code, and general problem-solving, leveraging reasoning patterns discovered through large-scale reinforcement learning. This model offers a 32K context length and demonstrates strong performance on benchmarks, making it suitable for applications requiring robust analytical capabilities.

Loading preview...

DeepSeek-R1-Distill-Qwen-7B: Reasoning-Optimized Language Model

This model is a 7.6 billion parameter distilled version of DeepSeek-R1, built upon the Qwen2.5-Math-7B architecture. Developed by DeepSeek-AI, it leverages advanced reasoning patterns derived from the larger DeepSeek-R1 model, which was trained using large-scale reinforcement learning (RL) without initial supervised fine-tuning (SFT) to foster complex chain-of-thought (CoT) capabilities.

Key Capabilities

  • Enhanced Reasoning: Inherits and distills sophisticated reasoning abilities from DeepSeek-R1, excelling in mathematical, coding, and general reasoning tasks.
  • Distillation Advantage: Demonstrates that reasoning patterns from larger, more complex models can be effectively transferred to smaller, dense models, achieving strong performance.
  • Competitive Benchmarks: Shows strong results on various benchmarks, including AIME 2024, MATH-500, GPQA Diamond, and LiveCodeBench, often outperforming other models in its size class.
  • Extended Context: Supports a context length of 32,768 tokens, allowing for processing longer inputs and generating more comprehensive responses.

Good For

  • Complex Problem Solving: Ideal for applications requiring detailed step-by-step reasoning, such as mathematical proofs, code generation, and logical puzzles.
  • Resource-Efficient Reasoning: Provides high-quality reasoning capabilities in a smaller, more deployable package compared to its larger DeepSeek-R1 counterpart.
  • Research and Development: Useful for researchers exploring distillation techniques and the transfer of advanced reasoning skills to more compact models.
  • Applications requiring robust analytical performance.