diff --git a/README.md b/README.md
index ee44ccb..4c18870 100644
--- a/README.md
+++ b/README.md
@@ -35,6 +35,7 @@ python -m logpilot.cli analyze <path> [OPTIONS]
 - `--recursive` : Recurse into subdirectories (directory mode)
 - `--include <glob>` : Glob pattern(s) to include (repeatable)
 - `--exclude <glob>` : Glob pattern(s) to exclude (repeatable)
+- `--model <model>` : LLM model to use (default: gpt-4)
 
 ### Example
 
@@ -62,6 +63,12 @@ Summarize a log file and write to a file:
 python -m logpilot.cli analyze mylog.txt --out-file summary.txt
 ```
 
+Analyze logs with a different model:
+
+```sh
+python -m logpilot.cli analyze mylog.txt --model gpt-3.5-turbo
+```
+
 ## Project Structure
 
 - `logpilot/` - Core modules
diff --git a/logpilot/cli.py b/logpilot/cli.py
index 4d528cf..a64055a 100644
--- a/logpilot/cli.py
+++ b/logpilot/cli.py
@@ -71,6 +71,7 @@ def analyze(
         "--exclude",
         help="Glob pattern(s) to exclude (repeatable).",
     ),
+    model: str = typer.Option("gpt-4", help="LLM model to use (default: gpt-4)"),
 ):
     """Analyze logs using Copilot SDK."""
     import os
@@ -124,7 +125,7 @@ def analyze(
     if os.environ.get("LOGPILOT_MOCK_LLM") == "1":
         responses = ["Mocked summary: Something failed" for _ in prompts]
     else:
-        llm = LLMClient()
+        llm = LLMClient(model=model)
         responses = [llm.analyze(prompt) for prompt in prompts]
 
     # 6. Aggregate responses