whatmodel.sh / DeepSeek R1 Distill Qwen 7B

DeepSeek R1 Distill Qwen 7B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

DeepSeek R1 Distill Qwen 7B is an open-weights language model from DeepSeek, released January 2025. Its published size is 7.62B parameters and a 131k-token context window. Licence: mit.

48.3±0.9#376 of 719 LLMsfrom 31 benchmark results across 12 sources

$ how it compares

The LLMs ranked either side of DeepSeek R1 Distill Qwen 7B, out of 828 with a score, then more from DeepSeek.

#modelscorewhereRAMreleased
447Hermes 3 Llama 3.1 70B Nous Research49±1local~44GB2024-07
448Qwen2.5 7B Instruct Alibaba48±1api $0.30/M +local~6GB2024-09
449DeepSeek R1 Distill Qwen 7B DeepSeek48±1local~6GB2025-01
450XwinCoder-34B Xwin-LM48±1local~23GB2023-11
451StarChat2 15B v0.1 Hugging Face48±1local~12GB2024-03
more from DeepSeek
21DeepSeek V4.1 Flash DeepSeek78±1api $0.26/M +local~537GB2026-09
31DeepSeek V4 Pro (0813) DeepSeek76±1api $1.63/M +local~1010GB2026-08
35DeepSeek V3.2 Speciale DeepSeek76±3comparison only

see all LLMs · find the ones that fit your machine

$ what it's good at

coding▁▁▁▃▆██▇▆▄▃▁ #428 of 458 · 15% of its score · 6 tests

Trails Llama 3.2 1B Instruct, SmolLM2 1.7B Instruct, Mistral 7B Instruct v0.1 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
LiveCodeBench (version unspecified) lab37.6% ✎▄····▄▄··▄█▄ beats 1/6medium · avg 41%▇▇ 5.9%at its level
LiveCodeBench v2 +150.1% ✎▄··█·▄···▄▄█ beats 1/7medium · avg 50%▇▇ 4.0%at its level · several runs
CodeReviewQA: change localisation (hard), invariant accuracy0.2% ✎█▂▁▁▂▂▂▃▅▅▅▇ beats 1/47medium · avg 36%▇ 2.1%sharp · at its level · noisy
CodeReviewQA: solution identification (hard), invariant accuracy0.1% ✎▃··▁▂▂▃▂█▃▄▂ beats 0/47medium · avg 54%▇ 1.9%sharp · at its level · noisy
CodeReviewQA: automated code refinement, exact match rate6.7% ✎▁▁▁▃▁▁▂▂▃█▅▄ beats 5/47hard · avg 33%▇ 1.4%at its level
CodeReviewQA: change type recognition, invariant accuracy1.0% ✎▁▃▁······▁▃█ beats 2/47easy · avg 67%▇ 0.0%blunt · at its level · noisy
math▁▂▃▅▆▅▄█▄▁▁▁ #185 of 398 · 47% of its score · 6 tests

Beats MiniMax-M2.1, DeepSeek V3 0324, Qwen3 235B-A22B (Apr 2025, hybrid) here, though they rank higher overall. Trails Phi-4-mini-reasoning 3.8B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
AIME 2024 lab +455.5% ✎▃·▃·▃▃···█·▃ beats 3/7medium · avg 53%▇▇▇▇▇▇ 16%sharp · at its level · small run · several runs
MATH-500 lab +492.8% ✎▄·▄▄·▄▄█···▄ beats 3/7very easy · avg 89%▇▇▇▇ 10%small run · several runs
MATH (subset unspecified)87.9% ✎▃·····▃···▅█ beats 0/6very easy · avg 90%▇▇▇ 7.9%typical
AIME 202533.3% ✎▁·▁▁▁▂▅▆█▇▅▁ beats 20/133medium · avg 38%▇▇▇ 6.8%sharp · at its level · small run
OMEGA (math generalisation)28.5% ✎▄······█▄▄·█ beats 0/6hard · avg 30%▇▇ 4.6%at its level
THINK-Bench final-answer accuracy (reasoning models)68.5% ✎▄█·▄····▄▄·█ beats 1/7easy · avg 76%▇ 1.8%at its level
other tests6 not in a task · 37% of its score
testresultvs the others in its rundifficultycountswhy
GPQA Diamond lab +449.1% ✎▄·····█·█▄▄▄ beats 2/7hard · avg 50%▇▇▇▇▇▇▇▇▇▇▇ 29%sharp · at its level · small run · several runs
BRIDGE clinical text, zero-shot average (87 tasks)25.3% ✎▁▁▁▁▁▂▃▃▄▆█▃ beats 19/97hard · avg 32%▇ 3.3%blunt · at its level
SurgiQ surgical MCQ (zero-shot, log-probability scoring)34.0% ✎▂···▃▂·▃▂▃██ beats 1/17medium · avg 53%▇ 1.5%at its level
Open PT-LLM Leaderboard average (9 Portuguese tasks)50.4% ✎▃▁▁▁▁▁▁▃▂█▅▁ beats 69/276medium · avg 73%▇ 1.5%at its level · noisy
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy49.0% ✎▃·▁▁▂▁▃▅█▇▄▁ beats 26/100medium · avg 68%▇ 1.2%at its level · noisy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy53.9% ✎▁▁▁▁▁▁▁▃▄█▄▃ beats 24/146easy · avg 83%▇ 0.8%at its level · noisy

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run DeepSeek R1 Distill Qwen 7B · needs ~4.7 GB · from huggingface.co/lmstudio-community
    chat in the terminal
    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)
    llama serve -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run DeepSeek R1 Distill Qwen 7B · needs ~4.7 GB · from huggingface.co/lmstudio-community

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-7B-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run DeepSeek R1 Distill Qwen 7B · needs ~5.07 GB · from ollama.com
    chat in the terminal
    ollama run deepseek-r1:7b

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run DeepSeek R1 Distill Qwen 7B · needs ~4.3 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/DeepSeek-R1-Distill-Qwen-7B-4bit

Weights and documentation: huggingface.co/deepseek-ai ↗

$ how the score was worked out · open-thoughts/OpenThinker2-7B… 26%, deepseek-ai/DeepSeek-R1 model… 18%, allenai/Olmo-3-7B-Think model… 12% and 9 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
open-thoughts/OpenThinker2-7B model card5▇▇▇▇▇▇▇▇ 26%Apache-2.0
deepseek-ai/DeepSeek-R1 model card · lab-reported4▇▇▇▇▇ 18%mit
allenai/Olmo-3-7B-Think model card2▇▇▇▇ 12%Apache-2.0
Microsoft — Phi-4-Mini Technical Report3▇▇▇ 11%CC BY 4.0
microsoft/Phi-4-mini-reasoning model card3▇▇▇ 10%MIT
open-thoughts/OpenThinker-7B model card4▇▇ 7%Apache-2.0
CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models (arXiv:2503.16167v2)4▇▇ 5%CC-BY-4.0
BRIDGE Medical Leaderboard (Harvard Medical School / Brigham and Women's YLab)1▇ 3%MIT
EuroEval leaderboards (results collection), English generative leaderboard2▇ 2%MIT
THINK-Bench: Evaluating Thinking Efficiency and Chain-of-Thought Quality of Large Reasoning Models (arXiv:2505.22113v1)1▇ 2%CC-BY-4.0
SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models (arXiv:2606.08071v1)1▇ 2%CC BY 4.0
Open Portuguese LLM Leaderboard (archived)1▇ 1%Apache-2.0
prior · pull toward the middle▇ 0%
weighted average= 48.3
$ quantization evidence · 21 measurements

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.

precisiontestfull → quantizedkept
AWQ 4-bit (AutoAWQ; vLLM)AIME 2024 accuracy (mean of 3 passes; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.467 → 0.46699.8%source ↗
AWQ 4-bit (AutoAWQ; vLLM)Average of AIME 2024, FOLIO and Temporal accuracy (published 'Avg'; 3 passes each; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.668 → 0.65798.4%source ↗
FP8 W8A8 dynamic (FP8 weights symmetric per-channel, FP8 activations dynamic per-token; llm-compressor FP8_DYNAMIC)AIME 2024 pass@10.532 → 0.532100.0%source ↗
FP8 W8A8 dynamic (FP8 weights symmetric per-channel, FP8 activations dynamic per-token; llm-compressor FP8_DYNAMIC)MATH-500 pass@10.937 → 0.936100.0%source ↗
FP8 W8A8 dynamic (FP8 weights symmetric per-channel, FP8 activations dynamic per-token; llm-compressor FP8_DYNAMIC)Reasoning average of 3 tasks (AIME 2024 pass@1, MATH-500 pass@1, GPQA Diamond pass@1)0.658 → 0.65799.8%source ↗
FP8 W8A8 dynamic (FP8 weights symmetric per-channel, FP8 activations dynamic per-token; llm-compressor FP8_DYNAMIC)GPQA Diamond pass@10.505 → 0.50399.5%source ↗
FP8 W8A8 dynamic (FP8 weights symmetric per-channel, FP8 activations dynamic per-token; llm-compressor FP8_DYNAMIC)IFEval 0-shot instruction-level strict acc (OpenLLM v2 config, chat template)0.394 → 0.3999.1%source ↗
GPTQ 3-bit (vLLM)Average of AIME 2024, FOLIO and Temporal accuracy (published 'Avg'; 3 passes each; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.668 → 0.38457.5%source ↗
GPTQ 3-bit (vLLM)AIME 2024 accuracy (mean of 3 passes; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.467 → 0.17838.1%source ↗
GPTQ 4-bit (vLLM)Average of AIME 2024, FOLIO and Temporal accuracy (published 'Avg'; 3 passes each; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.668 → 0.60790.9%source ↗
GPTQ 4-bit (vLLM)AIME 2024 accuracy (mean of 3 passes; vLLM, T=0.6, top-p 0.95, 32768 max tokens)0.467 → 0.38983.3%source ↗
INT4 W4A16 (weights only, symmetric per-group, group size 128; card states GPTQ via llm-compressor)IFEval 0-shot instruction-level strict acc (OpenLLM v2 config, chat template)0.394 → 0.394100.0%source ↗
INT4 W4A16 (weights only, symmetric per-group, group size 128; card states GPTQ via llm-compressor)MATH-500 pass@10.937 → 0.93399.6%source ↗
INT4 W4A16 (weights only, symmetric per-group, group size 128; card states GPTQ via llm-compressor)GPQA Diamond pass@10.505 → 0.49898.6%source ↗
INT4 W4A16 (weights only, symmetric per-group, group size 128; card states GPTQ via llm-compressor)Reasoning average of 3 tasks (AIME 2024 pass@1, MATH-500 pass@1, GPQA Diamond pass@1)0.658 → 0.64798.3%source ↗
INT4 W4A16 (weights only, symmetric per-group, group size 128; card states GPTQ via llm-compressor)AIME 2024 pass@10.532 → 0.50995.7%source ↗
INT8 W8A8 (weights symmetric per-channel, activations symmetric per-token; SmoothQuant + GPTQ via llm-compressor)AIME 2024 pass@10.532 → 0.552103.8%source ↗
INT8 W8A8 (weights symmetric per-channel, activations symmetric per-token; SmoothQuant + GPTQ via llm-compressor)IFEval 0-shot instruction-level strict acc (OpenLLM v2 config, chat template)0.394 → 0.402102.1%source ↗
INT8 W8A8 (weights symmetric per-channel, activations symmetric per-token; SmoothQuant + GPTQ via llm-compressor)Reasoning average of 3 tasks (AIME 2024 pass@1, MATH-500 pass@1, GPQA Diamond pass@1)0.658 → 0.663100.7%source ↗
INT8 W8A8 (weights symmetric per-channel, activations symmetric per-token; SmoothQuant + GPTQ via llm-compressor)GPQA Diamond pass@10.505 → 0.506100.2%source ↗
INT8 W8A8 (weights symmetric per-channel, activations symmetric per-token; SmoothQuant + GPTQ via llm-compressor)MATH-500 pass@10.937 → 0.9399.3%source ↗

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json