n8n workflows and the 2026–27 AI model shutdowns

Short answer: n8n saves the model ID inside every AI node, and several of n8n's own defaults are on the shutdown list: the OpenAI Chat Model defaults to gpt-5-mini (shuts down 11 December 2026), Anthropic Chat Model 1.3 defaults to claude-sonnet-4-5-20250929 (30 November 2026), and the OpenAI node's transcribe and translate operations hardcode whisper-1 (26 February 2027). Export your workflows, scan them, and change the model in each node it lists.

Scan your n8n workflows for retiring models (free, in your browser)

Where the model is set

Node	Where the model is	Default
OpenAI Chat Model 1–1.1	`parameters.model` (text)	`gpt-5-mini` (`gpt-4o-mini` before n8n 2.5.0)
OpenAI Chat Model 1.2–1.3	`parameters.model.value`	`gpt-5-mini` (`gpt-4.1-mini` before n8n 2.5.0)
OpenAI node: message a model	`parameters.modelId.value`	none preselected
OpenAI node: generate audio	`parameters.model` (only tts-1 or tts-1-hd)	`tts-1`
OpenAI node: transcribe, translate	hardcoded in n8n, no setting	`whisper-1`
Anthropic Chat Model 1.3	`parameters.model.value`	`claude-sonnet-4-5-20250929`
Anthropic Chat Model 1.4+	`parameters.model.value`	`claude-sonnet-4-6` or later
OpenAI (legacy) node	`parameters.model` / `chatModel`	`gpt-3.5-turbo` (already shut down)
HTTP Request	wherever you typed it in the body	–

n8n writes a plain setting into the workflow JSON only when it differs from the default, so a node left on its default model has no model in the file. The scanner knows these defaults and reports them as "node default". n8n changed the OpenAI Chat Model default from gpt-4o-mini to gpt-5-mini in version 2.5.0 (January 2026) without a new node version.

Fix

Export every workflow: n8n export:workflow --all --output=workflows.json on self-hosted n8n, or Download from each workflow's menu, or GET /api/v1/workflows with your X-N8N-API-KEY.
Paste the file into the scanner. It lists each workflow and node with the model, the shutdown date and the replacement.
Open each listed node and pick the replacement model from the list. For nodes on their default, picking any model saves it into the workflow, so a later n8n default change can no longer move it.
AI Agent with tools: if you move to gpt-6-luna, use OpenAI Chat Model 1.3, which sends requests through the Responses API by default. Versions up to 1.2 use Chat Completions, where OpenAI allows tool calls on gpt-6-luna only with reasoning effort none, and n8n does not show the reasoning effort option for gpt-6 models.
Transcribe or translate audio: the OpenAI node calls whisper-1 with no way to change it. Before 26 February 2027, move these steps to an HTTP Request node that calls OpenAI with a current transcription model, or update n8n if a newer release changes the operation.
Run each changed workflow once with real input and compare the output before the deadline.

Many workflows or services to move? The AI Model Shutdown Kit patches every n8n workflow through the API with a backup of each, scans code and config, and compares the old and new model on your own prompts (answers, tool calls, tokens, cost) before you switch. 18-page guide included.

Get the kit ($29, agencies $79)

Watch out

Removing temperature matters for gpt-6-luna when reasoning effort is not none; n8n sends it only if you added it under Options.
Updating a published workflow through the API re-publishes it unless you pass publishIfActive: false.
Make sure Anthropic Chat Model nodes are not on claude-sonnet-4-5-20250929 before 30 November 2026.

Sources

Other platforms

AI model shutdown calendar and scanner

Facts checked 28 October 2026. Not affiliated with OpenAI, Anthropic, Google or n8n.

Did this fix it?