diff --git a/api/app/clients/BaseClient.js b/api/app/clients/BaseClient.js
index 2d008b9991..8f931f8a5e 100644
--- a/api/app/clients/BaseClient.js
+++ b/api/app/clients/BaseClient.js
@@ -797,7 +797,8 @@ class BaseClient {
           promptTokens,
           completionTokens,
           balance: balanceConfig,
-          model: responseMessage.model,
+          /** Note: When using agents, responseMessage.model is the agent ID, not the model */
+          model: this.model,
           messageId: this.responseMessageId,
         });
       }
diff --git a/api/app/clients/specs/BaseClient.test.js b/api/app/clients/specs/BaseClient.test.js
index 15328af644..f13c9979ac 100644
--- a/api/app/clients/specs/BaseClient.test.js
+++ b/api/app/clients/specs/BaseClient.test.js
@@ -821,6 +821,56 @@ describe('BaseClient', () => {
     });
   });
 
+  describe('recordTokenUsage model assignment', () => {
+    test('should pass this.model to recordTokenUsage, not the agent ID from responseMessage.model', async () => {
+      const actualModel = 'claude-opus-4-5';
+      const agentId = 'agent_p5Z_IU6EIxBoqn1BoqLBp';
+
+      TestClient.model = actualModel;
+      TestClient.options.endpoint = 'agents';
+      TestClient.options.agent = { id: agentId };
+
+      TestClient.getTokenCountForResponse = jest.fn().mockReturnValue(50);
+      TestClient.recordTokenUsage = jest.fn().mockResolvedValue(undefined);
+      TestClient.buildMessages.mockReturnValue({
+        prompt: [],
+        tokenCountMap: { res: 50 },
+      });
+
+      await TestClient.sendMessage('Hello', {});
+
+      expect(TestClient.recordTokenUsage).toHaveBeenCalledWith(
+        expect.objectContaining({
+          model: actualModel,
+        }),
+      );
+
+      const callArgs = TestClient.recordTokenUsage.mock.calls[0][0];
+      expect(callArgs.model).not.toBe(agentId);
+    });
+
+    test('should pass this.model even when this.model differs from modelOptions.model', async () => {
+      const instanceModel = 'gpt-4o';
+      TestClient.model = instanceModel;
+      TestClient.modelOptions = { model: 'gpt-4o-mini' };
+
+      TestClient.getTokenCountForResponse = jest.fn().mockReturnValue(50);
+      TestClient.recordTokenUsage = jest.fn().mockResolvedValue(undefined);
+      TestClient.buildMessages.mockReturnValue({
+        prompt: [],
+        tokenCountMap: { res: 50 },
+      });
+
+      await TestClient.sendMessage('Hello', {});
+
+      expect(TestClient.recordTokenUsage).toHaveBeenCalledWith(
+        expect.objectContaining({
+          model: instanceModel,
+        }),
+      );
+    });
+  });
+
   describe('getMessagesWithinTokenLimit with instructions', () => {
     test('should always include instructions when present', async () => {
       TestClient.maxContextTokens = 50;
diff --git a/api/server/services/Endpoints/agents/initialize.js b/api/server/services/Endpoints/agents/initialize.js
index fd2e42511d..e71270ef85 100644
--- a/api/server/services/Endpoints/agents/initialize.js
+++ b/api/server/services/Endpoints/agents/initialize.js
@@ -306,6 +306,7 @@ const initializeClient = async ({ req, res, signal, endpointOption }) => {
       }
     } catch (err) {
       logger.error(`[initializeClient] Error processing agent ${agentId}:`, err);
+      skippedAgentIds.add(agentId);
     }
   }
 
@@ -315,7 +316,12 @@ const initializeClient = async ({ req, res, signal, endpointOption }) => {
       if (checkAgentInit(agentId)) {
         continue;
       }
-      await processAgent(agentId);
+      try {
+        await processAgent(agentId);
+      } catch (err) {
+        logger.error(`[initializeClient] Error processing chain agent ${agentId}:`, err);
+        skippedAgentIds.add(agentId);
+      }
     }
     const chain = await createSequentialChainEdges([primaryConfig.id].concat(agent_ids), '{convo}');
     collectEdges(chain);