Feature Request: Add LiteLLM as a Global Model Provider

[flefevre]

Summary

I would like to request support for LiteLLM as a global model provider in Langflow, alongside the existing providers such as Ollama, IBM, and OpenAI.

This would allow users to configure LiteLLM centrally in the Global Model Providers section and use it across flows, including for building Knowledge Bases powered by models served through LiteLLM.

Background

Langflow recently introduced Global Model Providers, allowing language model providers to be configured centrally and reused across the entire Langflow instance.

Instead of managing provider credentials and settings inside individual LLM components, users can now:

• Configure providers once in the Models pane
• Reuse them across flows and components
• Avoid credential sprawl and configuration drift
• Rotate keys or switch providers without manually editing every flow

Currently supported providers include:

• Ollama
• IBM
• OpenAI

However, LiteLLM is not yet available as a global model provider.

Motivation

LiteLLM is widely used as a unified gateway to multiple model backends (OpenAI-compatible APIs, Azure, Anthropic, local models, etc.). Supporting LiteLLM as a global provider would:

• Enable centralized configuration of LiteLLM endpoints
• Allow organizations to standardize on LiteLLM as their model gateway
• Simplify integration with self-hosted or proxied LLM infrastructure
• Enable Knowledge Base creation using models served via LiteLLM
• Improve enterprise and multi-provider workflows

This is especially valuable for teams using LiteLLM as an abstraction layer to manage cost control, logging, routing, or multi-model strategies.

Proposed Enhancement

Add LiteLLM as a selectable provider in the Global Model Providers configuration, with support for:

• Base URL configuration
• API key configuration
• Model selection
• Optional custom headers (if applicable)
• Compatibility with existing LLM components and Knowledge Base features

Ideally, LiteLLM would be usable anywhere other providers (OpenAI, Ollama, IBM) are currently supported, including:

• Chat models
• Embeddings (if supported through LiteLLM)
• Knowledge Base ingestion and querying

Expected Impact

• Greater flexibility in deployment architectures
• Easier enterprise adoption
• Reduced configuration complexity for multi-model environments
• Improved interoperability with OpenAI-compatible APIs

Thank you for considering this feature request!

<img width="1052" height="801" alt="image" src="https://github.com/user-attachments/assets/923666c3-982e-4a1e-8724-c4609b3e5a39" />

[ParisNeo]

This is a very interesting feature to add.