DeepSeek R1 Distill Qwen 14B
Intelligence
15.8
#321/520
Coding
—
Math
55.7
#130/265
Speed
—
Pricing
—
DeepSeek-R1 is the first-generation reasoning model built atop DeepSeek-V3 (671B total parameters, 37B activated per token). It incorporates large-scale reinforcement learning (RL) to enhance its chain-of-thought and reasoning capabilities, delivering strong performance in math, code, and multi-step reasoning tasks.
DeepSeek R1 Distill Qwen 14B is DeepSeek’s offering for advanced natural language processing tasks. It features an Intelligence Index of 15.8 and a Math Index of 55.7, making it suitable for applications requiring complex reasoning and coding capabilities. The model is set to be released on January 20, 2025.
When to Use DeepSeek R1 Distill Qwen 14B
✓ Best For
- Advanced natural language understanding
- Mathematical problem solving
- Coding assistance
✗ Not Ideal For
- Basic conversational tasks
- High-speed processing requirements
How DeepSeek R1 Distill Qwen 14B Compares
Intelligence Index · Higher is better
AnthropicTII UAEDeepSeekInclusionAIAlibaba
Benchmark Profile
Math Index
GoogleAlibabaDeepSeekNVIDIA
Intelligence · Coding · Math
Intelligence
Coding
Math
All Benchmark Scores (14)
| Benchmark | Score | |
|---|---|---|
| Intelligence Index | 15.8 | |
| Math Index | 55.7 | |
| MMLU-Pro | 74% | |
| GPQA | 484% | |
| LiveCodeBench | 376% | |
| HLE | 44% | |
| SciCode | 23.9% | |
| IFBench | 22.1% | |
| LCR | 7% | |
| AIME | 66.7% | |
| AIME 2025 | 55.7% | |
| MATH 500 | 94.9% | |
| AIME 2024 | 80% | |
| MATH-500 | 93.9% |
Data: Artificial Analysis · Updated: April 2, 2026
Frequently Asked Questions (15)
When was DeepSeek R1 Distill Qwen 14B released?
DeepSeek R1 Distill Qwen 14B was released on January 20, 2025.
Who created DeepSeek R1 Distill Qwen 14B?
DeepSeek R1 Distill Qwen 14B was created by DeepSeek.
How intelligent is DeepSeek R1 Distill Qwen 14B?
DeepSeek R1 Distill Qwen 14B scores 16 (estimated) on the Artificial Analysis Intelligence Index, placing it above average among other open weight models of similar size (median: 15).
How verbose is DeepSeek R1 Distill Qwen 14B?
When evaluated on the Intelligence Index, DeepSeek R1 Distill Qwen 14B generated 12M output tokens, which is better than average compared to other open weight models of similar size (median: 20M).
Is DeepSeek R1 Distill Qwen 14B a reasoning model?
Yes, DeepSeek R1 Distill Qwen 14B is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer.
What input modalities does DeepSeek R1 Distill Qwen 14B support?
DeepSeek R1 Distill Qwen 14B supports text input.
What output modalities does DeepSeek R1 Distill Qwen 14B support?
DeepSeek R1 Distill Qwen 14B supports text output.
Can DeepSeek R1 Distill Qwen 14B process images?
No, DeepSeek R1 Distill Qwen 14B does not support image input. It can only process text.
Is DeepSeek R1 Distill Qwen 14B multimodal?
No, DeepSeek R1 Distill Qwen 14B is not multimodal. It only supports text input.
What is the context window of DeepSeek R1 Distill Qwen 14B?
DeepSeek R1 Distill Qwen 14B has a context window of 130k tokens. This determines how much text and conversation history the model can process in a single request.
Is DeepSeek R1 Distill Qwen 14B open source?
Yes, DeepSeek R1 Distill Qwen 14B is open weights. The model weights are publicly available and can be downloaded for self-hosting.
How many parameters does DeepSeek R1 Distill Qwen 14B have?
DeepSeek R1 Distill Qwen 14B has 14 billion parameters.
What is the license for DeepSeek R1 Distill Qwen 14B?
DeepSeek R1 Distill Qwen 14B is released under the Apache 2.0 license. This license allows commercial use.
How does DeepSeek R1 Distill Qwen 14B perform on benchmarks?
DeepSeek R1 Distill Qwen 14B achieves a score of 16 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
Is DeepSeek R1 Distill Qwen 14B available via API?
DeepSeek R1 Distill Qwen 14B is an open weights model that can be self-hosted.
