[GH-ISSUE #20813] feat: add ability to define custom models with config files #19295

New Issue

GiteaMirror · 2026-04-20T01:42:25-05:00

GiteaMirror commented

2026-04-20 01:42:25 -05:00

Originally created by @t18n on GitHub (Jan 20, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/20813

Check Existing Issues

I have searched for all existing open AND closed issues and discussions for similar requests. I have found none that is comparable to my request.

Verify Feature Scope

I have read through and understood the scope definition for feature requests in the Issues section. I believe my feature request meets the definition and belongs in the Issues section instead of the Discussions.

Problem Description

Open WebUI’s model system is powerful, but it lacks effective tooling for large-scale management. The current UI is not intuitive for managing a large number of custom models.

I would prefer a code-driven workflow. For example, I want to define and manage models programmatically—generate or modify 50+ custom models for different use cases via an AI agent or scripts, restart the server, and have all model definitions in /models automatically loaded, validated, and ready for testing and refinement.

This approach would allow me to quickly tweak configurations, create and maintain variants of assistants (e.g., Senior Programmer – gpt-5.2, Senior Programmer – opus), and efficiently update or delete models with proper version control. The current UI-based workflow does not support this kind of scalable, reproducible model management.

Desired Solution you'd like

Dual model sources (UI + filesystem):
Open WebUI loads models from the existing source (database / UI‑managed) and additionally from a /models directory containing JSON model definitions. Both sources can be used together. File‑based models are loaded on app/container startup and, ideally, support dynamic reload without restart.
Declarative, version‑controlled models:
Self‑hosters can manage models as JSON files in /models, enabling automation, fast iteration, and version control, while UI‑managed models continue to work unchanged.
UI compatibility and clarity:
File‑managed models appear in the UI alongside UI‑created ones, clearly marked as config‑managed, and are fully usable in chat without forcing users away from the existing UI workflow.

Alternatives Considered

Base model abstraction with runtime swapping:
Allow defining a logical assistant (e.g., “Senior Programmer”) with a default base model (e.g., GPT‑5.2). In chat, users can optionally swap the underlying base model (e.g., Opus) without duplicating assistant definitions.

Additional Context

No response

Originally created by @t18n on GitHub (Jan 20, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/20813 ### Check Existing Issues - [x] I have searched for all existing **open AND closed** issues and discussions for similar requests. I have found none that is comparable to my request. ### Verify Feature Scope - [x] I have read through and understood the scope definition for feature requests in the Issues section. I believe my feature request meets the definition and belongs in the Issues section instead of the Discussions. ### Problem Description Open WebUI’s model system is powerful, but it lacks effective tooling for large-scale management. The current UI is not intuitive for managing a large number of custom models. I would prefer a code-driven workflow. For example, I want to define and manage models programmatically—generate or modify 50+ custom models for different use cases via an AI agent or scripts, restart the server, and have all model definitions in /models automatically loaded, validated, and ready for testing and refinement. This approach would allow me to quickly tweak configurations, create and maintain variants of assistants (e.g., Senior Programmer – gpt-5.2, Senior Programmer – opus), and efficiently update or delete models with proper version control. The current UI-based workflow does not support this kind of scalable, reproducible model management. ### Desired Solution you'd like - **Dual model sources (UI + filesystem):** Open WebUI loads models from the existing source (database / UI‑managed) *and* additionally from a `/models` directory containing JSON model definitions. Both sources can be used together. File‑based models are loaded on app/container startup and, ideally, support dynamic reload without restart. - **Declarative, version‑controlled models:** Self‑hosters can manage models as JSON files in `/models`, enabling automation, fast iteration, and version control, while UI‑managed models continue to work unchanged. - **UI compatibility and clarity:** File‑managed models appear in the UI alongside UI‑created ones, clearly marked as config‑managed, and are fully usable in chat without forcing users away from the existing UI workflow. ### Alternatives Considered - **Base model abstraction with runtime swapping:** Allow defining a logical assistant (e.g., **“Senior Programmer”**) with a default base model (e.g., GPT‑5.2). In chat, users can optionally swap the underlying base model (e.g., Opus) without duplicating assistant definitions. ### Additional Context _No response_

GiteaMirror referenced this issue

2026-04-20 05:47:44 -05:00

[PR #19295] [CLOSED] feat: Admins setting custom default interface settings #25167

GiteaMirror referenced this issue

2026-04-25 13:13:47 -05:00

[PR #19295] [CLOSED] feat: Admins setting custom default interface settings #40797

GiteaMirror referenced this issue

2026-04-29 23:48:08 -05:00

[PR #19295] [CLOSED] feat: Admins setting custom default interface settings #48215

GiteaMirror referenced this issue

2026-05-06 09:17:21 -05:00

[PR #19295] [CLOSED] feat: Admins setting custom default interface settings #64023