diff --git a/README.md b/README.md
index d41ec8d..cf99e65 100644
--- a/README.md
+++ b/README.md
@@ -96,6 +96,7 @@ ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags baseline
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags ocr
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags virtualbox
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags docker
+ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags local_llm
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags grobid
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags languagetool
 ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags quarto
@@ -107,6 +108,49 @@ ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags desktop
 
 You can also combine tags, e.g. `--tags baseline,docker,grobid,vscode`.
 
+
+## Local LLMs (Ollama + Open WebUI)
+
+This repository includes a `local_llm` role that runs:
+
+- `ollama` (API on `127.0.0.1:11434`)
+- `open-webui` (web UI on `http://127.0.0.1:3000`)
+
+Both services run in Docker with persistent storage under `/opt/local-llm`:
+
+- Ollama models: `/opt/local-llm/ollama`
+- Open WebUI data: `/opt/local-llm/open-webui`
+
+Install only this role:
+
+```sh
+ansible-playbook -i inventory -K playbooks/lab-stack.yml --tags local_llm
+```
+
+Optional model preloading (to avoid manual `ollama pull` after install):
+
+```yaml
+# host_vars/localhost.yml
+local_llm_models:
+  - llama3.2:3b
+```
+
+Optional Open WebUI defaults:
+
+```yaml
+# host_vars/localhost.yml
+local_llm_webui_auth: true
+local_llm_webui_default_models: "llama3.2:3b"
+```
+
+Quick checks:
+
+```sh
+curl -s http://127.0.0.1:11434/api/tags | jq .
+curl -I http://127.0.0.1:3000
+docker ps --format 'table {{.Names}}\t{{.Status}}\t{{.Ports}}' | grep -E 'ollama|open-webui'
+```
+
 ## Day-to-day
 
 ### Update software and configuration
diff --git a/playbooks/lab-stack.yml b/playbooks/lab-stack.yml
index ce70d9b..0a53e91 100644
--- a/playbooks/lab-stack.yml
+++ b/playbooks/lab-stack.yml
@@ -20,6 +20,9 @@
     - role: docker
       tags: [docker]
 
+    - role: local_llm
+      tags: [local_llm]
+
     - role: grobid
       tags: [grobid]
 
diff --git a/roles/local_llm/defaults/main.yml b/roles/local_llm/defaults/main.yml
new file mode 100644
index 0000000..abd5fd8
--- /dev/null
+++ b/roles/local_llm/defaults/main.yml
@@ -0,0 +1,30 @@
+---
+local_llm_docker_packages:
+  - docker-ce
+  - docker-ce-cli
+  - containerd.io
+  - docker-buildx-plugin
+  - docker-compose-plugin
+local_llm_python_packages:
+  - python3-docker
+
+local_llm_data_dir: /opt/local-llm
+local_llm_network_name: local-llm
+
+local_llm_ollama_image: ollama/ollama:latest
+local_llm_ollama_container_name: ollama
+local_llm_ollama_port: 11434
+local_llm_ollama_models_dir: "{{ local_llm_data_dir }}/ollama"
+
+local_llm_webui_image: ghcr.io/open-webui/open-webui:main
+local_llm_webui_container_name: open-webui
+local_llm_webui_port: 3000
+local_llm_webui_data_dir: "{{ local_llm_data_dir }}/open-webui"
+local_llm_webui_auth: true
+local_llm_webui_default_models: ""
+
+# Keep empty by default to avoid pulling large models unintentionally.
+# Example:
+# local_llm_models:
+#   - llama3.2:3b
+local_llm_models: []
diff --git a/roles/local_llm/tasks/main.yml b/roles/local_llm/tasks/main.yml
new file mode 100644
index 0000000..0401901
--- /dev/null
+++ b/roles/local_llm/tasks/main.yml
@@ -0,0 +1,77 @@
+---
+- name: Install Docker Engine packages
+  ansible.builtin.dnf:
+    name: "{{ local_llm_docker_packages }}"
+    state: present
+    update_cache: true
+
+- name: Install Python Docker SDK dependency for Ansible modules
+  ansible.builtin.dnf:
+    name: "{{ local_llm_python_packages }}"
+    state: present
+
+- name: Enable and start Docker service
+  ansible.builtin.systemd:
+    name: docker
+    enabled: true
+    state: started
+
+- name: Create local LLM data directories
+  ansible.builtin.file:
+    path: "{{ item }}"
+    state: directory
+    owner: root
+    group: root
+    mode: "0755"
+  loop:
+    - "{{ local_llm_data_dir }}"
+    - "{{ local_llm_ollama_models_dir }}"
+    - "{{ local_llm_webui_data_dir }}"
+
+- name: Ensure dedicated Docker network exists
+  community.docker.docker_network:
+    name: "{{ local_llm_network_name }}"
+    state: present
+
+- name: Run Ollama container
+  community.docker.docker_container:
+    name: "{{ local_llm_ollama_container_name }}"
+    image: "{{ local_llm_ollama_image }}"
+    restart_policy: unless-stopped
+    state: started
+    recreate: true
+    published_ports:
+      - "127.0.0.1:{{ local_llm_ollama_port }}:11434"
+    volumes:
+      - "{{ local_llm_ollama_models_dir }}:/root/.ollama"
+    networks:
+      - name: "{{ local_llm_network_name }}"
+
+- name: Wait for Ollama API to be reachable
+  ansible.builtin.wait_for:
+    host: 127.0.0.1
+    port: "{{ local_llm_ollama_port }}"
+    timeout: 120
+
+- name: Run Open WebUI container
+  community.docker.docker_container:
+    name: "{{ local_llm_webui_container_name }}"
+    image: "{{ local_llm_webui_image }}"
+    restart_policy: unless-stopped
+    state: started
+    recreate: true
+    published_ports:
+      - "127.0.0.1:{{ local_llm_webui_port }}:8080"
+    volumes:
+      - "{{ local_llm_webui_data_dir }}:/app/backend/data"
+    env:
+      OLLAMA_BASE_URLS: "http://{{ local_llm_ollama_container_name }}:11434"
+      WEBUI_AUTH: "{{ local_llm_webui_auth | default('True') | string }}"
+      DEFAULT_MODELS: "{{ local_llm_webui_default_models | default('') | string }}"
+    networks:
+      - name: "{{ local_llm_network_name }}"
+
+- name: Pull configured Ollama models (optional)
+  ansible.builtin.command: "docker exec {{ local_llm_ollama_container_name }} ollama pull {{ item }}"
+  loop: "{{ local_llm_models }}"
+  changed_when: false