I have tried everything to get my API key to work and I can’t figure out where to go or who to talk to, to fix it. Support has dropped off the map and stopped responding to my messages. Please keep in mind I’m proficient at best when it comes to this stuff, so the more you talk to me like I’m 5 the better. Here’s a summery if everything I’ve tried as reported by Replit:
Issue:
Every call to /v1/chat/completions returns HTTP 429 with insufficient_quota, regardless of model. The /v1/models endpoint works fine (HTTP 200), confirming the API key is valid and authenticated.
What I’ve tried:
-
Verified billing is active — $50 in prepaid credits, $120/month budget limit, auto-recharge enabled.
-
Confirmed account tier — Tier 1.
-
Tested multiple models —
gpt-5.2,gpt-4o, andgpt-3.5-turboall return the same 429 error. Even the cheapest model fails. -
Tested the models endpoint —
GET /v1/modelsreturns 200 OK with a full list of models, proving the key authenticates correctly. -
Generated multiple new API keys — Created fresh
sk-proj-keys from the dashboard. Same error on every key. -
Adjusted spending limits — Set and re-set the monthly budget/limit in billing settings. No change.
-
Tested via curl and the official OpenAI Node.js SDK — Same result in both cases.
-
Contacted OpenAI support — Submitted a ticket but response times have been slow.
-
Provided a screen recording — Recorded a live test at support’s request showing the full HTTP request/response with headers, demonstrating the 429 on completions and 200 on models.
Key evidence:
-
GET /v1/models→ 200 OK (key is valid) -
POST /v1/chat/completions→ 429 insufficient_quota (on every model)
Conclusion:
The API key authenticates successfully but the billing/quota system does not recognize the paid credits for completions. This appears to be an account provisioning issue on OpenAI’s side, not a configuration problem on mine.
Environment: Node.js, official OpenAI SDK, running on Replit.f