LLMBridge

LLMBridge is a concept of a single platform where users can access and switch between any AI provider — cloud or self-hosted — without changing their application code. Just update the config and run one command.

It is built on top of Bifrost, an open-source high-performance AI gateway that handles provider routing, failover, and load balancing out of the box. Bifrost turns what would normally require separate integrations for each provider into a single, ready-made solution.

The meter billing infrastructure using Lago is in the lago branch — tracking token usage per user and enabling payment through local payment methods, removing the dependency on international cards.

---

Run

Clone the repo and cd into it:

git clone https://github.com/monirz/llmbridge.git
cd llmbridge

Then start all services:

docker compose up --build

Once everything is up:

• http://localhost:9000 — LLMBridge chat UI
• http://localhost:8080 — Bifrost gateway dashboard

By default it runs with the self-hosted qwen2.5:3b model via Ollama — no API key needed. Bifrost, Ollama, and LLMBridge are all managed under the same docker compose.

---

Switch providers

Create a .env file in the project root. See .env.example for all available variables.

Ollama — local, no API key (default):

MODEL=ollama/qwen2.5:3b

Gemini:

GEMINI_API_KEY=your_key_here
MODEL=gemini/gemini-2.5-flash

Claude:

ANTHROPIC_API_KEY=your_key_here
MODEL=anthropic/claude-haiku-3-5

OpenAI:

OPENAI_API_KEY=your_key_here
MODEL=openai/gpt-4o-mini

Ollama starts automatically when MODEL is set to ollama/* or not set at all. Any other model skips Ollama and the download entirely.

---