From 222a131cf9679cf933327904e06f1e0debc6e207 Mon Sep 17 00:00:00 2001
From: anastasiia <anastasiya1155@gmail.com>
Date: Tue, 23 Apr 2024 17:39:31 +0100
Subject: [PATCH] update the model from gpt-4-1106-preview to gpt-4-turbo

---
 server/bleep/src/agent/model.rs      | 2 +-
 server/bleep/src/llm/call.rs         | 2 +-
 server/bleep/src/webserver/studio.rs | 8 ++++----
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/server/bleep/src/agent/model.rs b/server/bleep/src/agent/model.rs
index 917941e931..3ac0c249b7 100644
--- a/server/bleep/src/agent/model.rs
+++ b/server/bleep/src/agent/model.rs
@@ -41,7 +41,7 @@ const HEADROOM_CORRECTION: usize = GPT_4_TURBO_MAX_TOKENS - ACTUAL_MAX_TOKENS;
 // PS: when we want to fully utilize the model max context window, the correction is 0
 pub const GPT_4_TURBO_24K: LLMModel = LLMModel {
     tokenizer: "gpt-4-1106-preview",
-    model_name: "gpt-4-1106-preview",
+    model_name: "gpt-4-turbo",
     answer_headroom: 1024 + HEADROOM_CORRECTION,
     prompt_headroom: 2500 + HEADROOM_CORRECTION,
     history_headroom: 2048 + HEADROOM_CORRECTION,
diff --git a/server/bleep/src/llm/call.rs b/server/bleep/src/llm/call.rs
index 02231ddef3..56aa658493 100644
--- a/server/bleep/src/llm/call.rs
+++ b/server/bleep/src/llm/call.rs
@@ -82,7 +82,7 @@ pub async fn llm_call(
 ) -> anyhow::Result<impl Stream<Item = Result<Delta, api::Error>>> {
     let model = match req.model.as_deref() {
         Some(model) => model.to_owned(),
-        None => "gpt-4-turbo-preview".into(),
+        None => "gpt-4-turbo".into(),
     };
 
     let builder = {
diff --git a/server/bleep/src/webserver/studio.rs b/server/bleep/src/webserver/studio.rs
index 82a7662582..d74e56c926 100644
--- a/server/bleep/src/webserver/studio.rs
+++ b/server/bleep/src/webserver/studio.rs
@@ -31,7 +31,7 @@ use crate::{
 
 mod diff;
 
-const LLM_GATEWAY_MODEL: &str = "gpt-4-1106-preview";
+const LLM_GATEWAY_MODEL: &str = "gpt-4-turbo";
 
 fn studio_not_found() -> Error {
     Error::not_found("unknown code studio ID")
@@ -491,7 +491,7 @@ async fn token_counts(
         })
         .collect::<Vec<_>>();
 
-    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-1106-preview").unwrap();
+    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-turbo").unwrap();
     let per_doc_file = stream::iter(doc_context)
         .map(|file| async {
             if file.hidden {
@@ -652,14 +652,14 @@ pub async fn get_doc_file_token_count(
         .map(|sr| sr.text)
         .collect::<String>();
 
-    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-1106-preview").unwrap();
+    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-turbo").unwrap();
     let token_count = core_bpe.encode_ordinary(&content).len();
 
     Ok(Json(token_count))
 }
 
 fn count_tokens_for_file(path: &str, body: &str, ranges: &[Range<usize>]) -> usize {
-    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-1106-preview").unwrap();
+    let core_bpe = tiktoken_rs::get_bpe_from_model("gpt-4-turbo").unwrap();
 
     let mut chunks = Vec::new();