Hi, new here!
Am developing an app to help coders. I’m integrating different providers. Gemini is easy since they do provide free keys with generous credits daily, making testing easy.
I couldn’t find similar thing with OpenAPI api. Is there a provision like Gemini in openapi for builders like me? If not what’s the best way?
Not entirely free, but if you have an API account with positive balance it is possible that you might be eligible for daily complimentary tokens on traffic shared with openai.
You can read more here:
https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai
Thanks. I do not have an account yet with paid balance. But If there’s no free playground, I guess I will start with other models and come back to this later
Hi @evstevemd - you can test for free on a limited set of models including gpt-5-nano subject to limits (60 RPM, 200 RPD, 40,000 TPM). Give it a try and let me know if you have feedback.
Hi Richa,
How do I do that exactly? I’m currently interested to make my little tool work with OpenAPI than care what model it is. So anything that guarantees that it will work with advanced models is fine.
If you have a link to docs or something let me know. As of now I assume just change model name and boom, it is free with those constraints. If so please confirm.
Thanks for replying
Yes, you would change the model name to gpt-5-nano. Follow the examples on this docs page.
Evidence is to the contrary:
gpt-5-nano FAIL! Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}
Please explain.
@_j: I tried again shortly after @Richa_Deshpande made this comment and it does actually work now. At least when creating a brand new account.
Three-year ChatGPT users that only just discovered the API may not find “create a different account” ideal…
Within that class of account holder, there’s also the “3 month $18 or $5 trial API credit expired” set, that didn’t even know that existed when creating a ChatGPT login.
I will test this scenario again and if that’s correct, I will also raise this again.
It’s kind of odd that we encounter this question rather frequently even though it should work without any issues.
@evstevemd can you confirm that you previously had an OpenAI account, maybe only for ChatGPT, before trying to use the API’s free tier?
I have an account on personal email and this is a different account we use at work.
I didn’t test it yet as there’s a bug in the tool we are fixing so I will test later tonight or tomorrow and let you know
I’m getting “Error: OpenAI request failed: Error: client error (429)”, which is I have exceeded my limit. So seems it is not free s dime.
Model: gpt-5-nano
It did not work!
It says I have exceeded quota
I screenshot a “start building” flow you might see in the platform.openai.com site in the post linked below that is supposed to get a new account some free uses of select models, an undocumented quantity or capability of “free use” except by casual mention on the forum.
I expect this is a button and final “free” offered and successful if you have a completely original API account that was just setup, never offered an API trial credit (as any ChatGPT signup would be given up to program termination over a year ago), a new profile not triggering many possible guards against exploitation beyond simply having a common phone number or active sign-in credentials from providers or a cookied-up browser.
Or simply if you look like a “lead”.
It may be then that it is only successfully set up if you continue through completely, adding a payment method anyway, one that is unique to the platform. Consider Stripe as payment processor is also aggregating cards and personal information and business info passed by OpenAI to be easily used against you for detection of prior engagement. “Free tests but be prepared and willing to buy credits”.
I didn’t add payment because it wanted to charge. Since it is not clear and I didn’t wanted to have surprise in the wallet, Instoped that.
When adding a payment card, it is “authorized”, tested that a $5 charge would not be refused, that there is a real working card connected to banking, which can have a temporary hold on funds. However, this will be reversed out by your bank when it is not actually processed as a payment. Perhaps this is the reluctance point you encountered?
Yeah, that’s the thing.
Thanks
Thank you for testing this! We are still working on providing a better free testing experience for API so I will look into this with the team.
Currently this is only available for new accounts in the Free Trial tier on https://platform.openai.com/settings/organization/billing/overview. Can you confirm by testing an API call like the one below with an API Key from an organization on the Free Tier?
curl “https://api.openai.com/v1/responses”
-H “Content-Type: application/json”
-H “Authorization: Bearer [your-api-key]
-d ‘{
“model”: “gpt-5-nano”,
“input”: “Write a one-sentence bedtime story about a unicorn.”
}’
I will. Right now am testing after unlocking the free tokens. I will come with solid feedback after fixing all issues (we are using language not supported by sdk so we write everything ourlselves).
One thing I noted is that, once free tokens get exhausted, you get billed automatically. made me wonder if there is an API for checking how much token have been consumed? That will make it easy to get early warning in the app as token nears its end and avoid user getting surprised
Quick answer to that http request though:
{
"id": "resp_0b691ecc0088c04f00695c90cbbc808190a5701f0f815d897b",
"object": "response",
"created_at": 1767674059,
"status": "completed",
"background": false,
"billing": {
"payer": "openai"
},
"completed_at": 1767674063,
"error": null,
"incomplete_details": null,
"instructions": null,
"max_output_tokens": null,
"max_tool_calls": null,
"model": "gpt-5-nano-2025-08-07",
"output": [
{
"id": "rs_0b691ecc0088c04f00695c90cc3fec819093c582a9c3f0bafd",
"type": "reasoning",
"summary": []
},
{
"id": "msg_0b691ecc0088c04f00695c90cefbf48190b49351386f1fdb0f",
"type": "message",
"status": "completed",
"content": [
{
"type": "output_text",
"annotations": [],
"logprobs": [],
"text": "A gentle unicorn drifted through a moonlit meadow, whispering a lullaby that wrapped the world in sleep."
}
],
"role": "assistant"
}
],
"parallel_tool_calls": true,
"previous_response_id": null,
"prompt_cache_key": null,
"prompt_cache_retention": null,
"reasoning": {
"effort": "medium",
"summary": null
},
"safety_identifier": null,
"service_tier": "default",
"store": true,
"temperature": 1.0,
"text": {
"format": {
"type": "text"
},
"verbosity": "medium"
},
"tool_choice": "auto",
"tools": [],
"top_logprobs": 0,
"top_p": 1.0,
"truncation": "disabled",
"usage": {
"input_tokens": 17,
"input_tokens_details": {
"cached_tokens": 0
},
"output_tokens": 477,
"output_tokens_details": {
"reasoning_tokens": 448
},
"total_tokens": 494
},
"user": null,
"metadata": {}
}