DeepSeek R1 Distill Qwen 32B

by DeepSeek

DeepSeek R1 Distill Qwen 32B is a 32-billion parameter model distilled from DeepSeek R1, delivering state-of-the-art reasoning for dense models. Benchmarks include 72.6% on AIME 2024, 94.3% on MATH-500, 62.1% on GPQA Diamond, and 57.2% on LiveCodeBench, results comparable to OpenAI o1-mini across reasoning tasks. Released in January 2025 under the MIT license, the model uses a 64K token context window and captures chain-of-thought reasoning patterns from the 671B R1 parent through knowledge distillation, offering practical inference speed for production deployments while maintaining strong reasoning performance. DeepSeek R1 Distill Qwen 32B sits in a strong spot within the distill family, pairing reasoning performance close to o1-mini with a dense, interpretable architecture that excels at mathematics, coding, and complex problem-solving for academic and research platforms.

Key info

Input
Output
Features
Context window
64K
Max output
32K

Benchmarks

Independent benchmark scores — composite indices for reasoning, coding, and math, plus individual eval scores where available.

Global rank#372 of 596 LLMs
TierEfficient
Output speed0 tok/s
First token0.00s
Intelligence Index11.0
Math Index63.0
Reasoning & knowledge
MMLU-Pro
74%
GPQA Diamond
62%
Humanity's Last Exam
5%
Long-context reasoning
8%
Coding
LiveCodeBench
27%
SciCode
38%
Math & instruction following
AIME 2025
63%
IFBench
23%

Available routes

No routes currently available — DeepSeek R1 Distill Qwen 32B isn't routed through the Opper gateway right now. It may return.

Contact us about this model →

Available models from DeepSeek

Start building with 700+ models

One API key. Every major provider. Up and running in minutes.

Get startedView Documentation
DeepSeek R1 Distill Qwen 32B by DeepSeek — not currently on Opper | Opper AI