From 022e345d68a1fb67a40e5adf172ba418814ca0e7 Mon Sep 17 00:00:00 2001
From: lly835 <lly835@163.com>
Date: Thu, 26 Feb 2026 01:18:32 +0800
Subject: [PATCH] =?UTF-8?q?fix:=20=E4=BF=AE=E5=A4=8D=E5=90=8C=E4=BC=9A?=
 =?UTF-8?q?=E8=AF=9D=E8=B7=A8=E6=A8=A1=E5=9E=8B=E7=BC=93=E5=AD=98=E9=94=AE?=
 =?UTF-8?q?=E4=B8=8D=E5=91=BD=E4=B8=AD=E9=97=AE=E9=A2=98?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

移除 prepareRequestBody 中非 default sessionId 时的 model 前缀拼接，
使同一会话内切换模型时能复用缓存，与注释中'弱化 model 依赖'的设计意图一致。

Closes #328
---
 src/providers/openai/codex-core.js | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/src/providers/openai/codex-core.js b/src/providers/openai/codex-core.js
index 3160484..1aced0d 100644
--- a/src/providers/openai/codex-core.js
+++ b/src/providers/openai/codex-core.js
@@ -310,13 +310,11 @@ export class CodexApiService {
         delete cleanedBody.metadata;
 
         // 生成会话缓存键
-        // 默认弱化 model 依赖，以提升同会话跨模型的缓存命中率
-        // 如果 sessionId 为 'default'，则必须加上 model 以提供基础隔离
+        // 弱化 model 依赖，以提升同会话跨模型的缓存命中率
+        // 仅当 sessionId 为 'default' 时加上 model 前缀，提供基础隔离
         let cacheKey = sessionId;
         if (sessionId === 'default') {
             cacheKey = `${model}-default`;
-        } else {
-            cacheKey = `${model}-${sessionId}`;
         }
         
         let cache = this.conversationCache.get(cacheKey);