[whatmodel.sh](../../README.md) / DeepSeek R1 Distill Qwen 7B
# DeepSeek R1 Distill Qwen 7B
DeepSeek · released 2025-01 · 7.62B params · mit
**score 47.9** ±0.8 · #410 of 732 · from 32 benchmark results across 13 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks DeepSeek R1 Distill Qwen 7B among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁●▅▇█▇▅▄▃▁ #439 of 469 15% 6
math ▁▁▃▅█▆●█▄▂▁▁ #192 of 431 48% 7
coding · #439 of 469 · 15% of its score
Trails [Llama 3.2 1B Instruct](llama-3-2-1b.md), [SmolLM2 1.7B Instruct](smollm2-1-7b.md), [Mistral 7B Instruct v0.1](mistral-7b-instruct-v0-1.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LiveCodeBench (version unspecifie… 37.6% ▄····●▄··▄█▄ beats 1/6 medium · avg 42% 5.5% at its level
LiveCodeBench v2 50.1% ▄··●·▄···▄▄█ beats 1/7 medium · avg 50% 3.5% at its level · several runs
CodeReviewQA: solution identifica… 0.1% ●··▁▂▂▃▂█▃▄▂ beats 0/47 medium · avg 51% 2.3% sharp · at its level · noisy
CodeReviewQA: change localisation… 0.2% ●▂▁▁▂▂▂▃▅▅▅▇ beats 1/47 hard · avg 33% 2.3% sharp · at its level · noisy
CodeReviewQA: automated code refi… 6.7% ▁▁▁●▁▁▂▂▃█▅▄ beats 5/47 hard · avg 32% 1.8% at its level
CodeReviewQA: change type recogni… 1.0% ▁●▁······▁▃█ beats 2/47 easy · avg 67% 0.1% blunt · at its level · noisy
math · #192 of 431 · 48% of its score
Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 0324](deepseek-v3-0324.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md) here, though they rank higher overall. Trails [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Phi-4-mini-reasoning 3.8B](phi-4-mini-reasoning.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
AIME 2024 (maker) 55.5% ▃·▃·▃●···█·▃ beats 3/7 medium · avg 54% 14% sharp · at its level · small run · seve…
MATH (subset unspecified) 87.9% ●·····▃···▅█ beats 0/6 very easy · avg 91% 8.9% typical
MATH-500 (maker) 92.8% ▄·▄▄·●▄█···▄ beats 3/7 very easy · avg 89% 8.6% small run · several runs
AIME 2025 33.3% ▁·▁▁▁▂●▆█▇▅▁ beats 20/133 medium · avg 38% 7.1% sharp · at its level · small run
OMEGA (math generalisation) 28.5% ●······█▄▄·█ beats 0/6 hard · avg 31% 4.3% at its level
GSM8K 91.5% ▃··▃··▃●·▃·█ beats 3/7 very easy · avg 93% 3.9% sharp · noisy
THINK-Bench final-answer accuracy… 68.5% ▄●·▄····▄▄·█ beats 1/7 easy · avg 76% 1.7% at its level
other tests (6)
test result vs the others in its run difficulty counts why
GPQA Diamond (maker) 49.1% ▄·····●·█▄▄▄ beats 2/7 medium · avg 52% 26% sharp · at its level · small run · seve…
BRIDGE clinical text, zero-shot a… 25.3% ▁▁▁▁▁▂●▃▄▆█▃ beats 19/97 hard · avg 31% 4.0% blunt · at its level
Open PT-LLM Leaderboard average (… 50.4% ▃▁▁▁▁▁▁●▂█▅▁ beats 69/276 medium · avg 72% 1.8% at its level · noisy
SurgiQ surgical MCQ (zero-shot, l… 34.0% ▂···●▂·▃▂▃██ beats 1/17 medium · avg 53% 1.7% at its level
EuroEval HellaSwag (English), tes… 49.0% ▃·▁▁▂▁●▅█▇▄▁ beats 26/100 medium · avg 67% 1.5% sharp · at its level · noisy
EuroEval Life in the UK (English … 53.9% ▁▁▁▁▁▁▁●▄█▄▃ beats 24/146 easy · avg 82% 1.0% at its level · noisy
source results share of the score
open-thoughts/OpenThinker2-7B model card 5 26% ▇▇▇
deepseek-ai/DeepSeek-R1 model card (sub… 4 15% ▇▇
allenai/Olmo-3-7B-Think model card 2 13% ▇▇
microsoft/Phi-4-mini-reasoning model ca… 3 9% ▇
Microsoft 3 9% ▇
CodeReviewQA: The Code Review Comprehen… 4 7% ▇
open-thoughts/OpenThinker-7B model card 4 6% ▇
BRIDGE Medical Leaderboard 1 4% ▇
Unsupervised Confidence Calibration for… 1 4% ▇
EuroEval leaderboards 2 3% ▇
Open Portuguese LLM Leaderboard 1 2% ▇
THINK-Bench: Evaluating Thinking Effici… 1 2% ▇
SurgiQ: A Large-Scale Multi-Domain Benc… 1 2% ▇
prior: pull toward the middle 0% ▇
weighted average = 47.9
Licences: open-thoughts/OpenThinker2-7B model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · deepseek-ai/DeepSeek-R1 model card ([mit](../../licenses/deepseek-ai--DeepSeek-R1.txt)) · allenai/Olmo-3-7B-Think model card ([Apache-2.0](../../licenses/allenai--Olmo-3-7B-Think.txt)) · microsoft/Phi-4-mini-reasoning model card ([MIT](../../licenses/card-microsoft-phi-4-mini-reasoning.txt)) · Microsoft ([CC BY 4.0](../../licenses/report-2503.01743v2.txt)) · CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · open-thoughts/OpenThinker-7B model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · BRIDGE Medical Leaderboard ([MIT](../../licenses/bridge-medical-leaderboard-MIT.txt)) · Unsupervised Confidence Calibration for Reasoning LLMs from a Single Generation ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · THINK-Bench: Evaluating Thinking Efficiency and Chain-of-Thought Quality of Large Reasoning Models ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run DeepSeek R1 Distill Qwen 7B (Q4_K_M, needs ~4.7 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF/tree/959510a4c8eff7bae27769e9232dca7d3d2b4ed3))
```sh
llama cli -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF:Q4_K_M
```
Other versions: Q8_0 (~8.1 GB) `llama cli -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run DeepSeek R1 Distill Qwen 7B (Q4_K_M, needs ~4.7 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF/tree/959510a4c8eff7bae27769e9232dca7d3d2b4ed3))
```sh
lms get https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~8.1 GB) `lms get https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF@Q8_0 -y` · 4bit (~4.3 GB, Mac only) `lms get https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit -y` · 8bit (~8.1 GB, Mac only) `lms get https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-7B-8bit -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run DeepSeek R1 Distill Qwen 7B (Q4_K_M, needs ~5.07 GB, from [ollama.com](https://ollama.com/library/deepseek-r1/tags))
```sh
ollama run deepseek-r1:7b
```
Other versions: Q8_0 (~9.64 GB) `ollama run deepseek-r1:7b-qwen-distill-q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run DeepSeek R1 Distill Qwen 7B (4bit, needs ~4.3 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit/tree/21848dbf533d2518a1ef895104820d5ee51317ea))
```sh
mlx_lm.chat --model mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit
```
Other versions: 8bit (~8.1 GB, Mac only) `mlx_lm.chat --model mlx-community/DeepSeek-R1-Distill-Qwen-7B-8bit`
Model repository and documentation: [huggingface.co/deepseek-ai](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-7B)
## Published speed measurements
Reference results for the hardware, engine and settings below. These are not verified speeds of the downloads above or predictions for your machine.
hardware precision speed engine and settings
Intel Arc Graphics 130V/140… q4_k_m 6.7 tok/s ollama (version not recorded) · l…
Intel Arc Graphics 130V/140… q4_k_m 6.67 tok/s ollama (version not recorded) · l…
RTX 2080 q4_k_m 64.27 tok/s ollama (version not recorded) · l…
RTX 2080 q4_k_m 64.3 tok/s ollama (version not recorded) · l…
Each speed links to its source, which has the full settings.
## Quantization evidence
How much of the full-precision result a quantized version kept, measured within one study. Rows from the same study share a table and aren't independent. They don't describe every download of that precision.
precision test full → quantized kept
AWQ 4-bit (AutoAWQ; vLLM) AIME 2024 accuracy (mean of 3 pas… 0.467 → 0.466 99.8%
AWQ 4-bit (AutoAWQ; vLLM) Average of AIME 2024, FOLIO and T… 0.668 → 0.657 98.4%
FP8 W8A8 dynamic (FP8 weights… AIME 2024 pass@1 0.532 → 0.532 100.0%
FP8 W8A8 dynamic (FP8 weights… MATH-500 pass@1 0.937 → 0.936 100.0%
FP8 W8A8 dynamic (FP8 weights… Reasoning average of 3 tasks (AIM… 0.658 → 0.657 99.8%
FP8 W8A8 dynamic (FP8 weights… GPQA Diamond pass@1 0.505 → 0.503 99.5%
FP8 W8A8 dynamic (FP8 weights… IFEval 0-shot instruction-level s… 0.394 → 0.39 99.1%
GPTQ 3-bit (vLLM) Average of AIME 2024, FOLIO and T… 0.668 → 0.384 57.5%
GPTQ 3-bit (vLLM) AIME 2024 accuracy (mean of 3 pas… 0.467 → 0.178 38.1%
GPTQ 4-bit (vLLM) Average of AIME 2024, FOLIO and T… 0.668 → 0.607 90.9%
GPTQ 4-bit (vLLM) AIME 2024 accuracy (mean of 3 pas… 0.467 → 0.389 83.3%
INT4 W4A16 (weights only, sym… IFEval 0-shot instruction-level s… 0.394 → 0.394 100.0%
INT4 W4A16 (weights only, sym… MATH-500 pass@1 0.937 → 0.933 99.6%
INT4 W4A16 (weights only, sym… GPQA Diamond pass@1 0.505 → 0.498 98.6%
INT4 W4A16 (weights only, sym… Reasoning average of 3 tasks (AIM… 0.658 → 0.647 98.3%
INT4 W4A16 (weights only, sym… AIME 2024 pass@1 0.532 → 0.509 95.7%
INT8 W8A8 (weights symmetric … AIME 2024 pass@1 0.532 → 0.552 103.8%
INT8 W8A8 (weights symmetric … IFEval 0-shot instruction-level s… 0.394 → 0.402 102.1%
INT8 W8A8 (weights symmetric … Reasoning average of 3 tasks (AIM… 0.658 → 0.663 100.7%
INT8 W8A8 (weights symmetric … GPQA Diamond pass@1 0.505 → 0.506 100.2%
INT8 W8A8 (weights symmetric … MATH-500 pass@1 0.937 → 0.93 99.3%
Each kept value links to its source.
[interactive version](https://whatmodel.sh/model/deepseek-r1-distill-qwen-7b) · [JSON](https://whatmodel.sh/api?format=json&all=1)