Run DeepSeek-R1 with Ollama on your Mac and provide web-side access through Open-WebUI.
1. Install Ollama
Ollama official:/
Ollama is a lightweight AI inference framework that supports running LLMs (large language models) locally. First, download and install Ollama.
mac native tool brew install ollama
$ brew install --cask ollama
Running `brew update --auto-update`...
==> Auto-updated Homebrew!
Updated 2 taps (homebrew/core and homebrew/cask).
==> New Formulae
bpmnlint gersemi i686-elf-grub kirimase org-formation rattler-index semver todoist x86_64-elf-grub
cf-terraforming globstar immich-go largetifftools ov rhai sequoia-sqv trdsql yoke
cloudfoundry-cli gotz kafkactl lazyjournal pivy rpds-py sttr typioca ytt
fortitude hishtory kapp mox punktf sdl3_ttf tml unciv
==> New Casks
candy-crisis font-winky-sans opera-air trae-cn
consul fuse-t pairpods ua-midi-control
focu macskk pareto-security veracrypt-fuse-t
font-sf-mono-nerd-font-ligaturized nvidia-nsight-compute trae
You have 13 outdated formulae and 1 outdated cask installed.
==> Downloading /ollama/ollama/releases/download/v0.5.13/
==> Downloading from /github-production-release-asset-2e65be/658928958/2dc24c17-0bc0-487a-92d1-0265efd65a14?X-Amz-Algorithm=AWS4-
############################################################################################################################################################### 100.0%
==> Installing Cask ollama
==> Moving App '' to '/Applications/'
==> Linking Binary 'ollama' to '/opt/homebrew/bin/ollama'
πΊ ollama was successfully installed!
Check whether Ollama is installed successfully. The version number will be displayed, such as: ollama version is 0.5.13
$ ollama --version
Warning: could not connect to a running Ollama instance
Warning: client version is 0.5.13
2. Download the model
Download the DeepSeek-R1 model
Model download address:/library/deepseek-r1
This command will automatically download the DeepSeek-R1 version 1.5B model and store it locally.
$ ollama pull deepseek-r1:7b
pulling manifest
pulling 96c415656d37... 100% ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ 4.7 GB
pulling 369ca498f347... 100% ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ 387 B
pulling 6e4c38e1172f... 100% ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ 1.1 KB
pulling f4d24e9138dd... 100% ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ 148 B
pulling 40fb844194b2... 100% ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ 487 B
verifying sha256 digest
writing manifest
success
3. Run locally DeepSeek-R1
After execution, the terminal will enter interactive mode and can directly enter text to talk to the model for dialogue.
# View downloaded models
$ ollama list
NAME ID SIZE MODIFIED
deepseek-r1:7b 0a8c26691023 4.7 GB 24 seconds ago
# Run the model
$ ollama run deepseek-r1:7b
>>> Send a message (/? for help)4. Run through Open-WebUI
Using the Web interface interaction model, Open-WebUI can be installed. The tool provides a user-friendly web front-end, making DeepSeek-R1 easier to use.
Clone Open-WebUI repository
$ git clone /open-webui/
Cloning into 'open-webui'...
remote: Enumerating objects: 91391, done.
remote: Counting objects: 100% (131/131), done.
remote: Compressing objects: 100% (74/74), done.
remote: Total 91391 (delta 70), reused 57 (delta 57), pack-reused 91260 (from 2)
Receiving objects: 100% (91391/91391), 177.81 MiB | 3.98 MiB/s, done.
Resolving deltas: 100% (60008/60008), done.
Updating files: 100% (4575/4575), done.
Start the Open-WebUI container
Install docker by mac. After the installation is completed, there will be a docker program in the application. Click to start
brew install --cask --appdir=/Applications docker
Start docker
docker run -d \
-p 3000:8080 \
--add-host=:host-gateway \
-v open-webui:/app/backend/data \
--name open-webui \
--restart always \
/open-webui/open-webui:main
Unable to find image '/open-webui/open-webui:main' locally
main: Pulling from open-webui/open-webui
d51c377d94da: Pull complete
987cac002684: Pull complete
076b75118273: Pull complete
157e623d2984: Pull complete
40d5353a5918: Pull complete
4f4fb700ef54: Pull complete
aebeb0b4e5d0: Pull complete
03f562834d64: Pull complete
dc0f62a912f5: Pull complete
93fdf9ebd111: Pull complete
596be9ce6130: Pull complete
07dc67f42781: Pull complete
7c2ef53b15e7: Pull complete
e5511c24fa69: Pull complete
69de4f91fd38: Pull complete
Digest: sha256:74fc3c741a5f3959c116dd5abc61e4b27d36d97dff83a247dbb4209ffde56372
Status: Downloaded newer image for /open-webui/open-webui:main
26b786db658d187c2b82256fcbf33102c8c10c25b1087393483272e53708908b
β’ -p 3000:8080: Map the container's port 8080 to the native port 3000;
β’ --add-host=:host-gateway: Allows containers to access the host network;
β’ -v open-webui:/app/backend/data: Mount the data storage directory and save the status and data of the container.
β’ --restart always: Ensure that the container runs automatically after restarting;
β’ /open-webui/open-webui:main: Pull the latest version of Open-WebUI image.
After running the container, accesshttp://localhost:3000You can access Open-WebUI.
#Stop the container
docker stop open-webui
#Delete container
docker rm open-webui
#Delete stored data
docker volume rm open-webui