diff --git a/README.md b/README.md
index 23599b3cf..e80ea4dfb 100644
--- a/README.md
+++ b/README.md
@@ -57,6 +57,25 @@ COUNCIL_MODELS = [
 CHAIRMAN_MODEL = "google/gemini-3-pro-preview"
 ```
 
+### 4. Custom Self-Hosted Models (Optional)
+
+You can also add custom self-hosted models (e.g., Ollama, vLLM) by configuring `CUSTOM_MODELS` in `backend/config.py`:
+
+```python
+CUSTOM_MODELS = {
+    "ollama/llama3": {
+        "api_url": "http://localhost:11434/v1/chat/completions",
+        "api_key": "ollama"  # Optional, defaults to "custom"
+    }
+}
+
+# Don't forget to add it to the council list!
+COUNCIL_MODELS = [
+    # ... other models
+    "ollama/llama3",
+]
+```
+
 ## Running the Application
 
 **Option 1: Use the start script**
diff --git a/backend/config.py b/backend/config.py
index a9cf7c473..2119ff48b 100644
--- a/backend/config.py
+++ b/backend/config.py
@@ -13,9 +13,19 @@
     "openai/gpt-5.1",
     "google/gemini-3-pro-preview",
     "anthropic/claude-sonnet-4.5",
-    "x-ai/grok-4",
+    "x-ai/grok-4"
 ]
 
+# Custom models configuration
+# Format: { "model_id": { "api_url": "...", "api_key": "..." } }
+CUSTOM_MODELS = {
+    # Example for a local Ollama model:
+    # "ollama/llama3": {
+    #     "api_url": "http://localhost:11434/v1/chat/completions",
+    #     "api_key": "ollama"  # key is often ignored by local servers but required by some clients
+    # }
+}
+
 # Chairman model - synthesizes final response
 CHAIRMAN_MODEL = "google/gemini-3-pro-preview"
 
diff --git a/backend/openrouter.py b/backend/openrouter.py
index 118fb0b73..8e9b6d3f0 100644
--- a/backend/openrouter.py
+++ b/backend/openrouter.py
@@ -2,7 +2,7 @@
 
 import httpx
 from typing import List, Dict, Any, Optional
-from .config import OPENROUTER_API_KEY, OPENROUTER_API_URL
+from .config import OPENROUTER_API_KEY, OPENROUTER_API_URL, CUSTOM_MODELS
 
 
 async def query_model(
@@ -11,18 +11,26 @@ async def query_model(
     timeout: float = 120.0
 ) -> Optional[Dict[str, Any]]:
     """
-    Query a single model via OpenRouter API.
+    Query a single model via OpenRouter API or a custom endpoint.
 
     Args:
-        model: OpenRouter model identifier (e.g., "openai/gpt-4o")
+        model: Model identifier (e.g., "openai/gpt-4o" or custom ID)
         messages: List of message dicts with 'role' and 'content'
         timeout: Request timeout in seconds
 
     Returns:
         Response dict with 'content' and optional 'reasoning_details', or None if failed
     """
+    if model in CUSTOM_MODELS:
+        config = CUSTOM_MODELS[model]
+        api_url = config["api_url"]
+        api_key = config.get("api_key", "custom")  # Default to dummy key if not provided
+    else:
+        api_url = OPENROUTER_API_URL
+        api_key = OPENROUTER_API_KEY
+
     headers = {
-        "Authorization": f"Bearer {OPENROUTER_API_KEY}",
+        "Authorization": f"Bearer {api_key}",
         "Content-Type": "application/json",
     }
 
@@ -34,7 +42,7 @@ async def query_model(
     try:
         async with httpx.AsyncClient(timeout=timeout) as client:
             response = await client.post(
-                OPENROUTER_API_URL,
+                api_url,
                 headers=headers,
                 json=payload
             )