I'm struggling to use gpt-6 model

[jaspisus]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

I’m unable to configure newer OpenAI models, such as gpt-6, in my current setup. I can only see 52 models, with the newest one being gpt-5.6. I’ve attempted to add it manually, but it fails. I’ve tried using the OpenAI plugin as well as Workflow Orchestration. I’m puzzled by the requirement for an additional API key, but even after creating one, it doesn’t resolve the issue. Consequently, I still can’t select gpt-6-sol in the LLM Node “Model” settings.

2. Additional context or comments

No response

[dosubot[bot]]

Hi @jaspisus! I'm Dosu and I'm helping the Dify team.

GPT-6 (Sol, Luna, Astra) isn't in the official OpenAI plugin yet, which is why your predefined list stops at GPT-5.6. The plugin's released manifest is still version 1.0.5 [1], and its model list ends at gpt-5.6/gpt-5.6-sol/gpt-5.6-terra/gpt-5.6-luna [2]. GPT-6 support is in “feat(openai): add GPT-6 models and refresh dependencies”, which is still open and not merged [3]. Once it merges and you update the plugin from the Plugins page in your workspace, GPT-6 Sol/Luna/Astra should appear as predefined models in the LLM Node.

This also explains why adding gpt-6-sol manually as a custom model fails. GPT-6 needs request handling that the released plugin lacks, such as sending max_completion_tokens instead of max_tokens and stripping sampling parameters like temperature/top_p when reasoning is enabled [3]. Without that logic, OpenAI's API rejects the request.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[DevChiniwala]

While you wait for the official plugin update, here's a working workaround:

As Dosu mentioned, the official OpenAI plugin hasn't merged GPT-6 support yet (PR #3918
). But you don't have to wait — you can use GPT-6 right now with the OpenAI-API-compatible provider:

Go to Settings → Model Provider → OpenAI-API-compatible
Set the API endpoint to https://api.openai.com/v1
Enter your OpenAI API key (must have GPT-6 access on your OpenAI account)
Add the model name manually as gpt-6-sol (or gpt-6-luna / gpt-6-astra)
Important config tweaks (this is why manual addition via the native OpenAI plugin fails):
Set the max tokens parameter name to max_completion_tokens (not max_tokens)
If you're using reasoning/thinking mode, disable the temperature and top_p parameters — OpenAI's API will reject requests that include sampling params alongside reasoning
Set model_type to llm

This bypasses the predefined model list entirely. The OpenAI-API-compatible provider sends raw requests, so as long as your parameters match what OpenAI expects, it works.

For the permanent fix: Keep an eye on PR #3918
. Once it's merged, just update the OpenAI plugin from your Plugins page and GPT-6 models will appear natively in the LLM Node dropdown.

[Icaro0310]

You don't need Dify's dropdown to list it. Go to Settings > Model Provider > OpenAI > Add model, pick LLM as the type, type gpt-6 into the model name field exactly as the API expects it, set the context size, save. If you're relying on the predefined list and gpt-6 isn't there, update the langgenius/openai plugin from the marketplace first, the list ships with the plugin version.

If the model saves fine but requests come back with model_not_found or a 404, that error is coming from OpenAI, not Dify. It means your API key's project doesn't have access to gpt-6 yet, so check the model in your OpenAI account before debugging the Dify side.

—
More of my work lives here: github.com/Icaro0310 · Website: icaro0310.github.io