diff --git a/api/server/middleware/abortMiddleware.js b/api/server/middleware/abortMiddleware.js
index feed002b0e..96db0b60d4 100644
--- a/api/server/middleware/abortMiddleware.js
+++ b/api/server/middleware/abortMiddleware.js
@@ -6,6 +6,7 @@ const {
   countTokens,
   GenerationJobManager,
   recordCollectedUsage,
+  getTransactionsConfig,
   sanitizeMessageForTransmit,
   buildAbortedResponseMetadata,
 } = require('@librechat/api');
@@ -59,6 +60,7 @@ const isAbortError = (error) => {
  * @param {Array<Object>} params.collectedUsage - Usage metadata from all models
  * @param {string} [params.fallbackModel] - Fallback model name if not in usage
  * @param {string} [params.messageId] - The response message ID for transaction correlation
+ * @param {AppConfig['transactions']} [params.transactions] - Resolved transactions config
  */
 async function spendCollectedUsage({
   userId,
@@ -66,6 +68,7 @@ async function spendCollectedUsage({
   collectedUsage,
   fallbackModel,
   messageId,
+  transactions,
 }) {
   if (!collectedUsage || collectedUsage.length === 0) {
     return;
@@ -85,6 +88,7 @@ async function spendCollectedUsage({
       context: 'abort',
       messageId,
       model: fallbackModel,
+      transactions,
     },
   );
 
@@ -149,6 +153,8 @@ async function abortMessage(req, res) {
     responseMessage.metadata = abortMetadata;
   }
 
+  const transactions = getTransactionsConfig(req.config);
+
   // Spend tokens for ALL models from collectedUsage (handles parallel agents/addedConvo)
   if (collectedUsage && collectedUsage.length > 0) {
     await spendCollectedUsage({
@@ -157,11 +163,12 @@ async function abortMessage(req, res) {
       collectedUsage,
       fallbackModel: jobData?.model,
       messageId: jobData?.responseMessageId,
+      transactions,
     });
   } else {
     // Fallback: no collected usage, use text-based token counting for primary model only
     await db.spendTokens(
-      { ...responseMessage, context: 'incomplete', user: userId },
+      { ...responseMessage, context: 'incomplete', user: userId, transactions },
       { promptTokens, completionTokens },
     );
   }
diff --git a/api/server/middleware/abortMiddleware.spec.js b/api/server/middleware/abortMiddleware.spec.js
index 06e434065a..20c4700421 100644
--- a/api/server/middleware/abortMiddleware.spec.js
+++ b/api/server/middleware/abortMiddleware.spec.js
@@ -19,6 +19,7 @@ const mockRecordCollectedUsage = jest
 
 const mockGetMultiplier = jest.fn().mockReturnValue(1);
 const mockGetCacheMultiplier = jest.fn().mockReturnValue(null);
+const mockGetTransactionsConfig = jest.fn().mockReturnValue({ enabled: false });
 
 jest.mock('@librechat/data-schemas', () => ({
   logger: {
@@ -37,7 +38,9 @@ jest.mock('@librechat/api', () => ({
     abortJob: jest.fn(),
   },
   recordCollectedUsage: mockRecordCollectedUsage,
+  getTransactionsConfig: (...args) => mockGetTransactionsConfig(...args),
   sanitizeMessageForTransmit: jest.fn((msg) => msg),
+  buildAbortedResponseMetadata: jest.fn().mockReturnValue(null),
 }));
 
 jest.mock('librechat-data-provider', () => ({
@@ -75,7 +78,9 @@ jest.mock('./abortRun', () => ({
 
 const { logger } = require('@librechat/data-schemas');
 const { sendError } = require('~/server/middleware/error');
-const { handleAbortError, spendCollectedUsage } = require('./abortMiddleware');
+const { GenerationJobManager } = require('@librechat/api');
+const db = require('~/models');
+const { handleAbort, handleAbortError, spendCollectedUsage } = require('./abortMiddleware');
 
 const buildAbortRequest = () => ({
   body: {
@@ -310,3 +315,107 @@ describe('abortMiddleware - handleAbortError', () => {
     expect(sendError).toHaveBeenCalledTimes(1);
   });
 });
+
+/**
+ * The transactions config is resolved from the request's app config and must reach
+ * every write path in this file. `createTransaction` reads `transactions` from the
+ * caller-supplied data, so an omitted value is indistinguishable from enabled and
+ * the write proceeds even when `transactions.enabled` is false.
+ */
+describe('abortMiddleware - transactions config', () => {
+  const buildJobData = () => ({
+    model: 'gpt-4',
+    responseMessageId: 'msg-123',
+    conversationId: 'convo-123',
+    endpoint: 'agents',
+    sender: 'AI',
+    promptTokens: 25,
+    userMessage: {
+      messageId: 'user-msg-123',
+      parentMessageId: 'parent-123',
+      conversationId: 'convo-123',
+      text: 'hello',
+    },
+  });
+
+  const buildReq = () => ({
+    body: { abortKey: 'convo-123:1', endpoint: 'agents' },
+    user: { id: 'user-123', email: 'user@example.com' },
+    config: { transactions: { enabled: false } },
+  });
+
+  const buildRes = () => ({
+    headersSent: false,
+    setHeader: jest.fn(),
+    send: jest.fn(),
+  });
+
+  beforeEach(() => {
+    jest.clearAllMocks();
+    mockGetTransactionsConfig.mockReturnValue({ enabled: false });
+    mockRecordCollectedUsage.mockResolvedValue({ input_tokens: 100, output_tokens: 50 });
+    db.getConvo.mockResolvedValue({ title: 'Test Chat' });
+  });
+
+  it('forwards transactions through spendCollectedUsage to recordCollectedUsage', async () => {
+    const collectedUsage = [{ input_tokens: 100, output_tokens: 50, model: 'gpt-4' }];
+
+    await spendCollectedUsage({
+      userId: 'user-123',
+      conversationId: 'convo-123',
+      collectedUsage,
+      fallbackModel: 'gpt-4',
+      transactions: { enabled: false },
+    });
+
+    expect(mockRecordCollectedUsage).toHaveBeenCalledTimes(1);
+    expect(mockRecordCollectedUsage).toHaveBeenCalledWith(
+      expect.any(Object),
+      expect.objectContaining({ context: 'abort', transactions: { enabled: false } }),
+    );
+  });
+
+  it('resolves the config from req and forwards it on the collected-usage path', async () => {
+    const collectedUsage = [{ input_tokens: 100, output_tokens: 50, model: 'gpt-4' }];
+    GenerationJobManager.abortJob.mockResolvedValue({
+      success: true,
+      jobData: buildJobData(),
+      content: [],
+      text: 'partial',
+      collectedUsage,
+    });
+
+    const req = buildReq();
+    await handleAbort()(req, buildRes());
+
+    expect(logger.error).not.toHaveBeenCalled();
+    expect(mockGetTransactionsConfig).toHaveBeenCalledWith(req.config);
+    expect(mockRecordCollectedUsage).toHaveBeenCalledTimes(1);
+    expect(mockRecordCollectedUsage).toHaveBeenCalledWith(
+      expect.any(Object),
+      expect.objectContaining({ context: 'abort', transactions: { enabled: false } }),
+    );
+  });
+
+  it('resolves the config from req and forwards it on the token-count fallback path', async () => {
+    GenerationJobManager.abortJob.mockResolvedValue({
+      success: true,
+      jobData: buildJobData(),
+      content: [],
+      text: 'partial',
+      collectedUsage: [],
+    });
+
+    const req = buildReq();
+    await handleAbort()(req, buildRes());
+
+    expect(logger.error).not.toHaveBeenCalled();
+    expect(mockGetTransactionsConfig).toHaveBeenCalledWith(req.config);
+    expect(mockRecordCollectedUsage).not.toHaveBeenCalled();
+    expect(mockSpendTokens).toHaveBeenCalledTimes(1);
+    expect(mockSpendTokens).toHaveBeenCalledWith(
+      expect.objectContaining({ context: 'incomplete', transactions: { enabled: false } }),
+      expect.any(Object),
+    );
+  });
+});