Is GPT-5 not available in Assistants?
I have added GPT-5 to the allowed models in my project, but it’s not showing up in Assistants.
Same here, yet the model page for GPT-5 shows Assistants is a supported endpoint!
1 Like
Error code: 400 - {'error': {'message': "The requested model 'gpt-5-mini' cannot be used with the Assistants API.", 'type': 'invalid_request_error', 'param': 'model', 'code': 'unsupported_model'}}
Error code: 400 - {'error': {'message': "The requested model 'gpt-5' cannot be used with the Assistants API.", 'type': 'invalid_request_error', 'param': 'model', 'code': 'unsupported_model'}}
Apparently someone realizes “we will announce deprecation” means the endpoint is already deprecated, and will eat all your chat data stored in threads when it is shut off.
2 Likes
We don’t need fast releases of models. We need releases in which everything works smoothly, and what you are committing should work at least.
2 Likes
We really need this, hope to hear from the team on this issue.
2 Likes
I really hope, this is a temporary issue.
1 Like
The assistants API is de facto deprecated. I think “move to responses” already is the message they are sending? (you are going to need to do that eventually so perhaps now is the time?)
1 Like
There is a difference between deprecated and going to be deprecated in 2026.
It is not deprecated now, so it is supposed to work.
I think you just found out what they really meant.
I am also seeing the same with all 3 models “The requested model ‘gpt-5’ cannot be used with the Assistants API.”
1 Like
Same here, no ChatGPT 5 in the assistants and an error saying “‘gpt-5’ cannot be used with the Assistants API." when trying to create one using the API.
The doc clearly state that
We're working to achieve full feature parity between the Assistants and the Responses API, including support for Assistant-like and Thread-like objects and the Code Interpreter tool. When complete, we plan to formally announce the deprecation of the Assistants API with a target sunset date in the first half of 2026.
Upon deprecation, we will provide a clear migration guide from the Assistants API to the Responses API that allows developers to preserve all their data and migrate their applications. Until we formally announce the deprecation, we'll continue delivering new models to the Assistants API.
Meaning that the Assistant API is not going anywhere for a few more month and not without a clear migration path AND full parity, which we do not have ATM with the Response API.
With that in mind, it is imperative that the new model is included in the Assistant endpoint. Of course any delay for technical reason is understandable, but for us to understand OpenAI need to COMMUNICATE to its customers, that’s rule N°1 of running a business..
7 Likes
Same… annoying. I have customers breathing down my neck.
And why the hell depricate the absolute best thing about the OpneAI API?
This is … 
4 Likes
Not sure if it was the same for the other models but Tier one is greyed out for gpt-5, idk if it was the same for the other models
Has anyone tried with Usage Tier 2 or higher?
Just want to chime in that I am experiencing this as well(model works fine on completions api, just not the assistant api).
Would love to see gpt5 added to assistants api! The lower levels of hallucination is something we’re looking forward to using.
1 Like
I’m tier 5, it’s not a tiers thing and Responses and Completions work fine. I was elated to see GPT-5 appear with Assistants endpoint support on the model page, especially as they stopped at 4.1 and neglected o3 and o4-mini, but at this point if feels like:
- They will get around to adding it to Assistants eventually, not a high priority.
- It’s a mistake that Assistants was marked as “supported”
Fingers crossed it’s the former.
2 Likes
Well, if they are planning to depricate the whole assistants API then why add GTP-5 to it.
I will be moving all our customers to gpt-oss on groq. And we will just build our own assistant like layer on top of that.
So. Play stupid games and win stupid prizes, is my comment to OpenAI.
2 Likes
I contacted OpenAI support, and their AI support agent told me the following:
Thank you for sharing the screenshots. I see that you have successfully added GPT-5 and related models to your project’s “Allowed models,” but they are not available for selection when creating or editing an Assistant. Here’s what you need to know:
- Even if GPT-5 is allowed in your project settings, model availability in the Assistants API is rolling out gradually. Not all users and regions have access at the same time.
- For most users, GPT-5 will become available in the Assistants API as part of the phased rollout. You may see it appear without additional action when your account is included.
- If you see GPT-5 available in Playground or API, but not in Assistants, this is expected until the rollout is complete. What to do next:
- Keep checking the model dropdown when creating or editing Assistants—once GPT-5 is available for your account and region, it will appear.
- No further setup is required on your end since you have already allowed the model. Let me know if you have more questions or need help with other OpenAI features!
It seems we’ll have to wait.
4 Likes
This has been passed on to OpenAI. Maybe we can get a, hopefully short, timeline for when this feature will be available.
2 Likes