From f797e90d79a926cd92ba835bbc9510c45f238fb5 Mon Sep 17 00:00:00 2001
From: Danny Avila <danny@librechat.ai>
Date: Tue, 22 Jul 2025 15:52:54 -0400
Subject: [PATCH] =?UTF-8?q?=F0=9F=94=80=20feat:=20`moonshotai/kimi`=20Cont?=
 =?UTF-8?q?ext=20and=20OpenRouter=20Endpoint=20Token=20Config=20(#8604)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* ✨ feat: Enhance agent initialization with endpoint token configuration and round max context tokens

* feat: recognize moonshot/kimi model context window

* chore: remove unused i18n key
---
 api/server/services/Endpoints/agents/agent.js |  4 +-
 .../services/Endpoints/custom/initialize.js   |  3 +-
 api/utils/tokens.js                           |  9 +++-
 api/utils/tokens.spec.js                      | 42 +++++++++++++++++++
 client/src/locales/en/translation.json        |  3 +-
 5 files changed, 55 insertions(+), 6 deletions(-)

diff --git a/api/server/services/Endpoints/agents/agent.js b/api/server/services/Endpoints/agents/agent.js
index 3af2170d5c..a64ce97e78 100644
--- a/api/server/services/Endpoints/agents/agent.js
+++ b/api/server/services/Endpoints/agents/agent.js
@@ -131,7 +131,7 @@ const initializeAgent = async ({
   );
   const agentMaxContextTokens = optionalChainWithEmptyCheck(
     maxContextTokens,
-    getModelMaxTokens(tokensModel, providerEndpointMap[provider]),
+    getModelMaxTokens(tokensModel, providerEndpointMap[provider], options.endpointTokenConfig),
     4096,
   );
 
@@ -191,7 +191,7 @@ const initializeAgent = async ({
     resendFiles,
     toolContextMap,
     useLegacyContent: !!options.useLegacyContent,
-    maxContextTokens: (agentMaxContextTokens - maxTokens) * 0.9,
+    maxContextTokens: Math.round((agentMaxContextTokens - maxTokens) * 0.9),
   };
 };
 
diff --git a/api/server/services/Endpoints/custom/initialize.js b/api/server/services/Endpoints/custom/initialize.js
index 58da9f10dd..d2b7549699 100644
--- a/api/server/services/Endpoints/custom/initialize.js
+++ b/api/server/services/Endpoints/custom/initialize.js
@@ -141,8 +141,9 @@ const initializeClient = async ({ req, res, endpointOption, optionsOnly, overrid
       const options = getOpenAIConfig(apiKey, clientOptions, endpoint);
       if (options != null) {
         options.useLegacyContent = true;
+        options.endpointTokenConfig = endpointTokenConfig;
       }
-      if (!customOptions.streamRate) {
+      if (!clientOptions.streamRate) {
         return options;
       }
       options.llmConfig.callbacks = [
diff --git a/api/utils/tokens.js b/api/utils/tokens.js
index a8363db182..a35f18a51b 100644
--- a/api/utils/tokens.js
+++ b/api/utils/tokens.js
@@ -226,7 +226,14 @@ const xAIModels = {
   'grok-4': 256000, // 256K context
 };
 
-const aggregateModels = { ...openAIModels, ...googleModels, ...bedrockModels, ...xAIModels };
+const aggregateModels = {
+  ...openAIModels,
+  ...googleModels,
+  ...bedrockModels,
+  ...xAIModels,
+  // misc.
+  kimi: 131000,
+};
 
 const maxTokensMap = {
   [EModelEndpoint.azureOpenAI]: openAIModels,
diff --git a/api/utils/tokens.spec.js b/api/utils/tokens.spec.js
index 8a2e8ed11d..2d4f051584 100644
--- a/api/utils/tokens.spec.js
+++ b/api/utils/tokens.spec.js
@@ -714,3 +714,45 @@ describe('Claude Model Tests', () => {
     });
   });
 });
+
+describe('Kimi Model Tests', () => {
+  describe('getModelMaxTokens', () => {
+    test('should return correct tokens for Kimi models', () => {
+      expect(getModelMaxTokens('kimi')).toBe(131000);
+      expect(getModelMaxTokens('kimi-k2')).toBe(131000);
+      expect(getModelMaxTokens('kimi-vl')).toBe(131000);
+    });
+
+    test('should return correct tokens for Kimi models with provider prefix', () => {
+      expect(getModelMaxTokens('moonshotai/kimi-k2')).toBe(131000);
+      expect(getModelMaxTokens('moonshotai/kimi')).toBe(131000);
+      expect(getModelMaxTokens('moonshotai/kimi-vl')).toBe(131000);
+    });
+
+    test('should handle partial matches for Kimi models', () => {
+      expect(getModelMaxTokens('kimi-k2-latest')).toBe(131000);
+      expect(getModelMaxTokens('kimi-vl-preview')).toBe(131000);
+      expect(getModelMaxTokens('kimi-2024')).toBe(131000);
+    });
+  });
+
+  describe('matchModelName', () => {
+    test('should match exact Kimi model names', () => {
+      expect(matchModelName('kimi')).toBe('kimi');
+      expect(matchModelName('kimi-k2')).toBe('kimi');
+      expect(matchModelName('kimi-vl')).toBe('kimi');
+    });
+
+    test('should match Kimi model variations with provider prefix', () => {
+      expect(matchModelName('moonshotai/kimi')).toBe('kimi');
+      expect(matchModelName('moonshotai/kimi-k2')).toBe('kimi');
+      expect(matchModelName('moonshotai/kimi-vl')).toBe('kimi');
+    });
+
+    test('should match Kimi model variations with suffixes', () => {
+      expect(matchModelName('kimi-k2-latest')).toBe('kimi');
+      expect(matchModelName('kimi-vl-preview')).toBe('kimi');
+      expect(matchModelName('kimi-2024')).toBe('kimi');
+    });
+  });
+});
diff --git a/client/src/locales/en/translation.json b/client/src/locales/en/translation.json
index def5f5a431..79d6bc4dcf 100644
--- a/client/src/locales/en/translation.json
+++ b/client/src/locales/en/translation.json
@@ -466,7 +466,6 @@
   "com_nav_send_message": "Send message",
   "com_nav_setting_account": "Account",
   "com_nav_setting_balance": "Balance",
-  "com_nav_setting_beta": "Beta features",
   "com_nav_setting_chat": "Chat",
   "com_nav_setting_data": "Data controls",
   "com_nav_setting_general": "General",
@@ -1090,4 +1089,4 @@
   "com_ui_yes": "Yes",
   "com_ui_zoom": "Zoom",
   "com_user_message": "You"
-}
\ No newline at end of file
+}