Language model selection

[DavideDelbianco]

Self Checks

• This is only for bug report, if you would like to ask a question, please head to Discussions.
• 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.

Dify version

1.5.0

Cloud or Self Hosted

Self Hosted (Docker), Cloud

Steps to reproduce

Add a second copy of an OpenAI model, with a different key
Open an LLM node

✔️ Expected Behavior

The selected model should be based on an unique, not its name

❌ Actual Behavior

[dosubot]

Based on the codebase, Dify is designed to use unique identifiers for model selection, combining elements like the provider name, plugin ID, and model name to distinguish between different model configurations—even if they share the same display name. This mechanism should allow you to add multiple OpenAI models with different keys and select them unambiguously in the UI and backend logic. For example, the ModelProviderFactory assigns unique identifiers by combining the plugin ID and provider name, and model schema caching uses a composite key including credentials, provider, and model name to ensure uniqueness [source].

However, your report suggests that in version 1.4.1, the selection UI or logic is still relying on the model name alone, which leads to ambiguity when two models have the same name but different keys. I did not find any existing issues or documentation addressing this specific bug in the repositories.

Given the architecture, this may be a UI bug or a misconfiguration where the unique identifier is not being surfaced or used for selection. You may want to check if the model configuration or selection logic in your deployment is using only the model property (the name) instead of the full unique identifier (which should include provider and credentials context). If you are comfortable with code, reviewing the relevant UI and backend logic for model selection may help pinpoint where the name is being used instead of the unique identifier.

If you need a workaround before a fix is available, ensure that each model's name is unique as a temporary measure, or check if there are updates or patches after 1.4.1 that address this behavior. If you need more detailed guidance on where to look in the code, let me know!

To reply, just mention my name @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}   

[DavideDelbianco]

Hey @dosu this graphical bug is still actual in 1.5.0 shouldn't you mark this issue as "bug" ?

[dosubot]

You're right—this is a valid bug. The frontend UI still doesn't distinguish between models with the same name and provider but different credentials, so they're visually indistinguishable in the dropdown. The backend supports unique identification, but the UI only uses the model name and provider for selection and display, which causes this persistent issue in 1.5.0 as well. There are no documented guidelines on labeling, but by open source standards and Dify's own structure, this should be marked as a "bug" for proper tracking and visibility.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}