Deprecated model-ID scanner
Paste code, a config file or an n8n workflow and find every model ID that’s retiring or already retired, with days left and the official replacement.
Runs in your browser. Nothing you paste is uploaded.
.py, .ts, .json, .env, n8n exports…). They're read in your browser; nothing is uploaded.
Scanning…
| Model ID | Status | Shutdown | Replacement | Lines |
|---|
Matched against 201 deprecated, retired and legacy models from the official OpenAI, Anthropic and Gemini API deprecation pages (data last changed 2026-11-02). Active models aren't listed.
How to use it
- Paste code, an
.envor config file, or an n8n/automation workflow export, or drop several files at once. - Read the list: retired models first (their requests already fail), then the soonest shutdowns, with the line numbers where each ID appears.
- Copy the result as a checklist for your issue tracker, and use the request migration linter to check each request against its new model.
How the scanner works
It finds every model-ID-shaped string in your text, normalizes platform spellings (models/gemini-2.0-flash, openai/gpt-4o, anthropic.claude-3-5-sonnet-20240620-v1:0, claude-sonnet-4-5@20250929), and looks each one up in our model retirement data. That data is parsed daily from the official OpenAI, Anthropic and Gemini API deprecation pages, and every date change is reviewed before it’s published.
An exact match means the ID appears in the provider’s table. An alias match means it’s a listed ID without its date suffix, for example gpt-4o for gpt-4o-2024-08-06. The official tables list the dated IDs, so the scanner flags the alias and labels it; check your provider’s docs to see which snapshot it points to. Dates are for each provider’s own API; Amazon Bedrock and Vertex AI publish their own schedules.
It can’t see model names built at runtime (for example from environment variables), so also search your deployment config.
FAQ
How do I find which deprecated OpenAI or Claude models my code uses?
Paste your code or config here, or drop the files. The scanner finds every model ID and checks it against the official OpenAI, Anthropic and Gemini API deprecation pages. Nothing is uploaded.
Does it work with n8n, Make or Zapier workflows?
Yes for any export that contains model IDs as text, such as n8n workflow JSON. Paste the export and the scanner finds the model IDs in it.
Does it catch aliases like gpt-4o or claude-sonnet-4-5?
Yes. Providers’ tables often list dated snapshots (gpt-4o-2024-08-06, claude-sonnet-4-5-20250929). The scanner also flags the undated alias and labels it as an alias. It also understands models/, openai/, Amazon Bedrock (anthropic.…-v1:0) and Vertex AI (…@date) spellings.
Why isn't my active model flagged?
The scanner flags only models a provider has deprecated or retired. For active Claude models, Anthropic publishes a “not sooner than” date, which is the earliest it could retire them, so those stay off the list.
How is this different from a GitHub app or CLI scanner?
There’s nothing to install and no repository access: paste text or drop files and it runs in your browser.