Hello everyone,
I’m encountering a very persistent and perplexing 403 API error when trying to access gpt-3.5-turbo and gpt-4o-mini from my OpenAI project. I’m building a healthy eating app that relies on these models for meal suggestions.
The specific error I consistently receive is: 403 Project ‘proj_HBJ6ILJkgHIkVwd7M8Ug90LO’ does not have access to model ‘[model_name]’ This error occurs for both gpt-3.5-turbo and gpt-4o-mini.
I’ve already gone through extensive troubleshooting, trying to isolate the cause:
- OpenAI Account & Billing:
- I’ve meticulously checked my OpenAI account’s billing section (platform.openai.com/billing).
- Billing is active , and a valid payment method is linked and verified .
- There are no outstanding balances , expired free trial notifications, or any other apparent issues with my billing setup. My account is set to usage-based billing.
- API Keys:
- I’ve generated multiple new ‘You’ type secret API keys from the platform (platform.openai.com/api-keys).
- I’ve ensured these keys are copied precisely and used correctly.
- Testing Environments (confirming the error origin):
- Google Cloud Function (Node.js 20): My primary API calls are made from a Google Cloud Function. The function itself deploys successfully, receives requests, and its Google Cloud Logging shows that the 403 error is coming directly from the OpenAI API endpoint (my console.error logs confirm this).
- Local Node.js Script: To rule out any Cloud Function-specific issues, I’ve replicated the exact OpenAI API call (using the official openai Node.js library and dotenv for the key) on my local Windows machine. The result is the exact same 403 Project does not have access to model error , confirming the issue is not related to the Cloud Function’s environment, deployment, or the gcloud CLI itself.
- Models Attempted:
- Both gpt-3.5-turbo and gpt-4o-mini consistently return this specific 403 error.
Based on these thorough steps, I’ve ruled out:
- Incorrect API key (copy/paste issue or expired key).
- Google Cloud Function deployment/environment issues.
- Google Cloud project billing/permission issues.
Given that my billing is confirmed active, new keys have been generated, and the error persists across different execution environments and models, I’m at a loss for further debugging steps on my end.
Has anyone in the community encountered this specific “403 Project does not have access to model” error under similar circumstances (i.e., with active billing and valid keys)? Are there any non-obvious account settings, specific verification steps, or specific model access tiers that I might be missing for my project ID proj_HBJ…?
Any insights or suggestions from the community would be greatly appreciated.
Thank you!
