From f491dfd2431812402528e7777e95e02328992325 Mon Sep 17 00:00:00 2001
From: Kaspar Schleiser <kaspar@schleiser.de>
Date: Sun, 15 Mar 2026 21:34:23 +0100
Subject: [PATCH] refactor: use plain string model names

---
 src/cli.rs  |  5 +++--
 src/main.rs | 42 ++++++++++--------------------------------
 2 files changed, 13 insertions(+), 34 deletions(-)

diff --git a/src/cli.rs b/src/cli.rs
index 8c1ecbd..b9f6804 100644
--- a/src/cli.rs
+++ b/src/cli.rs
@@ -29,7 +29,7 @@ pub fn clap() -> clap::Command {
                 .env("OPENAI_KEY")
                 .hide_env_values(true)
                 .required(true)
-                .value_parser(clap::value_parser!(String))
+                .value_parser(value_parser!(String))
                 .num_args(1),
         )
         .arg(
@@ -38,7 +38,8 @@ pub fn clap() -> clap::Command {
                 .long("model")
                 .help("which LLM to use")
                 .env("PLEASE_MODEL")
-                .value_parser(value_parser!(super::Model)),
+                .default_value("gpt-5")
+                .value_parser(value_parser!(String)),
         )
         .arg(
             Arg::new("temp")
diff --git a/src/main.rs b/src/main.rs
index b475b58..0c3797a 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -3,7 +3,7 @@ use std::error::Error;
 use std::io::{stdout, Read, Write};
 
 use anyhow::anyhow;
-use clap::{ArgMatches, ValueEnum};
+use clap::ArgMatches;
 
 use openai::{
     chat::{ChatCompletion, ChatCompletionDelta, ChatCompletionMessage, ChatCompletionMessageRole},
@@ -13,23 +13,6 @@ use tokio::sync::mpsc::Receiver;
 
 mod cli;
 
-#[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, ValueEnum, Default)]
-pub(crate) enum Model {
-    /// Use gpt-3.5-turbo
-    #[default]
-    #[clap(name = "gpt-3.5-turbo", alias = "gpt-3")]
-    GPT3_5,
-    /// Use gpt-4
-    #[clap(name = "gpt-4")]
-    GPT4,
-    /// Use gpt-4o
-    #[clap(name = "gpt-4o")]
-    GPT4o,
-    /// Use gpt-4o-mini
-    #[clap(name = "gpt-4o-mini")]
-    GPT4oMini,
-}
-
 #[tokio::main]
 async fn main() {
     let matches = cli::clap().get_matches();
@@ -51,9 +34,7 @@ async fn main() {
 }
 
 async fn handle_task(matches: &ArgMatches) -> Result<(), Box<dyn Error>> {
-    let model: Model = matches
-        .get_one::<Model>("model")
-        .map_or_else(Model::default, |m| *m);
+    let model = matches.get_one::<String>("model").unwrap();
 
     let task: Vec<String> = matches
         .get_many("task")
@@ -69,7 +50,7 @@ async fn handle_task(matches: &ArgMatches) -> Result<(), Box<dyn Error>> {
             return Err(anyhow!("running on tty, no task given").into());
         }
         prompt = "You are an assistant returning Linux shell commands that accomplish the following task. Don't add explanations or notes.".to_string();
-        openai_request(model, &prompt, &task, temperature).await?;
+        openai_request(&model, &prompt, &task, temperature).await?;
     } else {
         if task.is_empty() {
             task.push_str("Please fix this.");
@@ -81,14 +62,14 @@ async fn handle_task(matches: &ArgMatches) -> Result<(), Box<dyn Error>> {
 
         std::io::stdin().read_to_string(&mut buffer).unwrap();
 
-        openai_request(model, &prompt, &buffer, temperature).await?;
+        openai_request(&model, &prompt, &buffer, temperature).await?;
     }
 
     Ok(())
 }
 
 async fn openai_request(
-    model: Model,
+    model: &str,
     prompt: &str,
     task: &str,
     temperature: f32,
@@ -113,14 +94,11 @@ async fn openai_request(
         tool_calls: Some(Vec::new()),
     });
 
-    let chat_stream = ChatCompletionDelta::builder(
-        model.to_possible_value().unwrap().get_name(),
-        messages.clone(),
-    )
-    .credentials(credentials)
-    .temperature(temperature)
-    .create_stream()
-    .await?;
+    let chat_stream = ChatCompletionDelta::builder(model, messages.clone())
+        .credentials(credentials)
+        .temperature(temperature)
+        .create_stream()
+        .await?;
 
     listen_for_tokens(chat_stream).await;
     Ok(())