Regression in 2.36.5 — Mistral Chat Model node fails with "Failed to parse URL from [object Request]" (works in 2.35.4)

Bug Description

Every call through the Mistral Cloud Chat Model node (@n8n/n8n-nodes-langchain.lmChatMistralCloud, typeVersion 1) fails immediately on n8n Cloud 2.36.5.

The same workflow, with the same API key and the same model, works correctly on a second n8n Cloud instance still running 2.35.4. This points to a regression introduced between 2.35.4 and 2.36.5.

Instance	Version	Mistral node
A	2.36.5 (n8n Cloud)	fails
B	2.35.4 (n8n Cloud)	works

Error returned by the node:

NodeOperationError: Unexpected HTTP client error:
TypeError: Failed to parse URL from [object Request]
cause: { name: "UnexpectedClientError" }

Key observations:

1. No HTTP request ever reaches Mistral. The failure occurs before the network call: the URL parser receives a Request object instead of a URL string.
2. The credential is valid and correctly loaded. The node's runtime options show mistral_api_key present, and the credential "Test connection" button succeeds (it issues GET /v1/models, which returns the full model catalogue).
3. The same API key works through the OpenAI Chat Model node (lmChatOpenAi, typeVersion 1.3) pointed at Mistral's OpenAI-compatible endpoint, in the same workflow and the same execution. Only the native Mistral node fails.
4. The model ID is valid. mistral-large-2512 is present in the live catalogue returned by the API for that key.
5. Not key-specific: tested with 11 separate Mistral API keys, with an identical failure on all of them.
6. Retries do not help. With maxRetries: 3 the node attempts 4 times, each failing identically after roughly 9 to 13 seconds.
7. Reproduced downstream of both chainLlm (v1.9) and agent (v3.1).
8. @langchain/core version reported in the node metadata: 1.2.0.

Side effect worth noting: when the node is set to onError: continueRegularOutput, the error is swallowed and passed downstream as { error: "..." }. The workflow then reports as "success" while producing no model output, which makes the fault easy to misdiagnose as a data problem rather than a node failure.

Possibly related, same error signature reported earlier against 2.5.x/2.6.x: #25318 (with #25292, #25323 and #25590 closed as duplicates). If that fix landed and has since regressed, this may be a reopening rather than a new defect.

To Reproduce

1. On n8n Cloud 2.36.5, create a workflow with a Manual Trigger and a Basic LLM Chain node.
2. Attach a Mistral Cloud Chat Model subnode, with a valid Mistral API key and the model mistral-large-2512.
3. Set the prompt to any static text, for example "Say hello in one sentence."
4. Execute the workflow.
5. The node fails after several seconds with: "Unexpected HTTP client error: TypeError: Failed to parse URL from [object Request]"

The identical workflow succeeds on n8n Cloud 2.35.4.

Expected behavior

The Mistral Cloud Chat Model node should send the request to the Mistral API and return the model's completion, as it does on 2.35.4 with the same key, the same model and the same workflow.

Debug Info

Debug info

core

• n8nVersion: 2.36.5
• platform: docker (cloud)
• nodeJsVersion: 24.18.1
• nodeEnv: production
• database: sqlite
• executionMode: regular
• concurrency: 5
• license: enterprise (sandbox)
• consumerId: 00000000-0000-0000-0000-000000000000

storage

• success: all
• error: all
• progress: false
• manual: true
• binaryMode: filesystem

pruning

• enabled: true
• maxAge: 168 hours
• maxCount: 2500 executions

client

• userAgent: mozilla/5.0 (windows nt 10.0; win64; x64) applewebkit/537.36 (khtml, like gecko) chrome/151.0.0.0 safari/537.36
• isTouchDevice: false

cluster

• instanceCount: 1
• versions: 2.36.5
• instances:
  • instanceKey: 6bd9d5bd-3f76-4a8f-a52d-74a9067d9104, hostId: main-walidatrag-n8n-7b6d9c4ff5-2jlmr, instanceType: main, instanceRole: leader, version: 2.36.5
• checks:
  • check: hostid-clash, status: succeeded, warnings: -
  • check: lifecycle, status: succeeded, warnings: -
  • check: split-brain, status: succeeded, warnings: -
  • check: version-mismatch, status: succeeded, warnings: -

Generated at: 2026-08-24T18:18:39.813Z

Operating System

n8n Cloud (managed)

n8n Version

2.36.5

Node.js Version

24.18.1

Database

PostgreSQL

Execution mode

main (default)

Hosting

n8n cloud