diff --git a/pkg/cli/cmd.go b/pkg/cli/cmd.go
index 088b9f4..427a192 100644
--- a/pkg/cli/cmd.go
+++ b/pkg/cli/cmd.go
@@ -11,12 +11,14 @@ import (
 
 var (
 	maxTokens int
+	model     string
 )
 
 func init() {
 	inputCmds := []*cobra.Command{newCmd, promptCmd, replyCmd}
 	for _, cmd := range inputCmds {
 		cmd.Flags().IntVar(&maxTokens, "length", config.OpenAI.DefaultMaxLength, "Max response length in tokens")
+		cmd.Flags().StringVar(&model, "model", config.OpenAI.DefaultModel, "The language model to use")
 	}
 
 	rootCmd.AddCommand(
@@ -274,7 +276,7 @@ var replyCmd = &cobra.Command{
 			response <- HandleDelayedResponse(receiver)
 		}()
 
-		err = CreateChatCompletionStream(config.OpenAI.DefaultModel, messages, maxTokens, receiver)
+		err = CreateChatCompletionStream(model, messages, maxTokens, receiver)
 		if err != nil {
 			Fatal("%v\n", err)
 		}
@@ -354,7 +356,7 @@ var newCmd = &cobra.Command{
 			response <- HandleDelayedResponse(receiver)
 		}()
 
-		err = CreateChatCompletionStream(config.OpenAI.DefaultModel, messages, maxTokens, receiver)
+		err = CreateChatCompletionStream(model, messages, maxTokens, receiver)
 		if err != nil {
 			Fatal("%v\n", err)
 		}
@@ -403,7 +405,7 @@ var promptCmd = &cobra.Command{
 
 		receiver := make(chan string)
 		go HandleDelayedResponse(receiver)
-		err := CreateChatCompletionStream(config.OpenAI.DefaultModel, messages, maxTokens, receiver)
+		err := CreateChatCompletionStream(model, messages, maxTokens, receiver)
 		if err != nil {
 			Fatal("%v\n", err)
 		}