Why did GPT-4 Turbo stop working for me

You're trying to select GPT-4 Turbo and it's either greyed out, missing from your model dropdown, or throwing errors when you try to use it. This hits ChatGPT Plus users who previously had access, and developers whose AP

Why did GPT-4 Turbo stop working for me

You're trying to select GPT-4 Turbo and it's either greyed out, missing from your model dropdown, or throwing errors when you try to use it. This hits ChatGPT Plus users who previously had access, and developers whose API calls suddenly started failing.

What's actually happening

GPT-4 Turbo isn't completely gone — OpenAI phased out certain versions and changed how the model appears in different interfaces. If you're a ChatGPT Plus user, you won't see "GPT-4 Turbo" as a standalone option anymore because OpenAI consolidated models under the standard "GPT-4" selector in late 2024. The interface now automatically routes to the most current GPT-4 variant, which includes the capabilities you used to get from the Turbo version.

For API users, the `gpt-4-turbo` endpoint still exists, but if you were using older snapshot versions like `gpt-4-turbo-preview` or `gpt-4-1106-preview`, those hit their deprecation dates. OpenAI announces these timelines months in advance, but if you hardcoded an old model name in your application, you'll get a 404 error stating the model doesn't exist. The current production endpoint is `gpt-4-turbo-2024-04-09` or the alias `gpt-4-turbo`.

Sometimes the issue isn't deprecation — it's your account status. If you've hit usage limits on a ChatGPT Plus or Team account, or if your API account ran out of credits or exceeded rate limits, GPT-4 Turbo becomes unavailable while GPT-3.5 remains accessible. You might see "GPT-4 currently unavailable" or the model selector won't let you switch from GPT-3.5.

How to fix it

For ChatGPT web users:

  • Log into chatgpt.com and look at the model selector at the top of a conversation
  • If you see only "GPT-4o" or "GPT-4" without a Turbo label, that's expected — you're already using the current GPT-4 architecture
  • Click your profile icon (bottom left), then Settings → Data controls → Check your Plus subscription status
  • If your subscription lapsed, you'll need to resubscribe at chatgpt.com/settings/billing to regain GPT-4 access
  • If you're hitting message limits (typically 40 messages per 3 hours for Plus users), you'll see a counter showing when your quota resets — wait it out or upgrade to Team for higher caps

For API developers:

  • Check your API calls — if you're using `gpt-4-turbo-preview` or any preview snapshot, update to `gpt-4-turbo` or `gpt-4-turbo-2024-04-09`
  • Log into platform.openai.com/api-keys and verify your API key is active and hasn't expired
  • Go to platform.openai.com/settings/organization/billing to confirm you have available credits — if your balance is $0 and you have no auto-recharge enabled, all GPT-4 requests fail
  • Check platform.openai.com/settings/organization/limits to see your tier — new accounts start at Tier 1 with low rate limits; if you're maxing out, you'll need to wait or request a tier increase by adding credits
  • Test with a simple curl command: `curl https://api.openai.com/v1/chat/completions -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" -d '{"model":"gpt-4-turbo","messages":[{"role":"user","content":"test"}]}'` — if this returns a 404, your model string is wrong; if it's a 429, you've hit rate limits

If you're still getting errors after checking these, look at the exact error message. "Model not found" means you're using a deprecated identifier. "Insufficient quota" means billing issues. "Rate limit exceeded" means you've maxed your tier allowance.

If that doesn't work

Screenshot the exact error message from the ChatGPT interface or your API response — include the full JSON error object if you're coding. Visit platform.openai.com/account/limits to check current usage against your tier caps. For ChatGPT web issues that persist after verifying your subscription, see how to contact OpenAI support with your account email and the date/time the problem started.

For model behaviour that seems wrong — different responses, formatting changes, refusing prompts it used to handle — that's a separate issue covered in our guide on GPT model behaviour changes. If nothing loads at all, check our ChatGPT not working troubleshooting article.

Questions people actually ask

Q: Will GPT-4 Turbo come back as a separate option?

A: No. OpenAI merged model variants under unified selectors. The GPT-4 option you see now uses the latest Turbo-class architecture — you're getting equivalent or better performance.

Q: How long do preview models stay active?

A: Typically 3-6 months after OpenAI releases the stable version. Always use the production endpoint like `gpt-4-turbo` instead of dated snapshots to avoid sudden breaks.

Q: Does this affect my existing conversations?

A: No. Past conversations stay intact. Only new messages route to current models.

What to remember

  • OpenAI consolidated GPT-4 variants — "GPT-4 Turbo" as a label disappeared but the technology didn't
  • API users must update from preview snapshots like `gpt-4-1106-preview` to `gpt-4-turbo`
  • Check your billing first — zero credits or lapsed subscriptions disable GPT-4 access immediately
  • Rate limits apply per tier on API accounts — platform.openai.com/settings/organization/limits shows your current usage
  • Different error messages mean different fixes — read the actual error code before assuming the model vanished

---

*openai-support.com is an independent resource, not affiliated with OpenAI Inc.*

Related help