# Supported LLM Providers for OpenMAIC: Complete Configuration Guide > Discover the eight supported LLM providers for OpenMAIC, including OpenAI, Anthropic, and Google Gemini. This guide details their configuration for seamless integration. - Repository: [MAIC/OpenMAIC](https://github.com/THU-MAIC/OpenMAIC) - Tags: how-to-guide - Published: 2026-09-09 --- **OpenMAIC supports eight LLM providers out-of-the-box—OpenAI, Anthropic, Azure OpenAI, Google Gemini, MiniMax, Qwen, Lemonade, and custom OpenAI-compatible endpoints—each configurable via environment variables and the central `providersConfig` object.** OpenMAIC is designed as a provider-neutral AI platform that abstracts LLM interactions through a unified configuration layer. The supported LLM providers for OpenMAIC are defined in the core TypeScript files and validated through automated tests, ensuring seamless switching between cloud APIs and local deployments. ## Built-in LLM Providers OpenMAIC ships with predefined adapters for major commercial and local LLM services. These providers are automatically registered in the `providersConfig` object and become available in the UI when their corresponding environment variables are detected. | Provider | Config ID | Description | |----------|-----------|-------------| | **OpenAI** | `openai` | Standard OpenAI API for GPT-4, GPT-4o, and GPT-5 series models | | **Anthropic** | `anthropic` | Claude family of models (Claude 3, 3.5, etc.) accessed via Anthropic's API | | **Azure OpenAI** | `azure-openai` | Microsoft Azure-hosted OpenAI endpoints with enterprise compliance | | **Google Gemini** | `gemini` | Google's Gemini model family including Flash and Pro variants | | **MiniMax** | `minimax` | MiniMax commercial LLM service including the MiniMax M3 model | | **Qwen** | `qwen` | Alibaba's Qwen series (Qwen 2.5, Qwen Max, etc.) | | **Lemonade** | `lemonade` | Local OpenAI-compatible mock server for offline development and testing | | **Custom** | user-defined | Any OpenAI-compatible API endpoint via manual `providersConfig` extension | ## Configuration Architecture Provider capabilities are centralized in [`lib/store/settings.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/lib/store/settings.ts), which exports the `providersConfig` TypeScript interface. This configuration object stores API keys, base URLs, and available model lists for each service. The system exposes two critical validation functions in the same file: - **`hasUsableLLMProvider()`** – Checks if at least one provider has valid credentials - **`isLLMProviderConfigured(providerId)`** – Verifies that a specific provider ID has non-empty API keys and reachable endpoints When the application initializes, [`lib/ai/providers.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/lib/ai/providers.ts) reads this configuration and constructs the appropriate request adapters for each enabled service. ## Enabling Providers via Environment Variables Each built-in provider maps to specific environment variables defined in `.env.example`. OpenMAIC uses these variables to populate the runtime `providersConfig` object. ```bash # OpenAI OPENAI_API_KEY=sk-... OPENAI_BASE_URL=https://api.openai.com/v1 # Anthropic (Claude) ANTHROPIC_API_KEY=sk-ant-... # Azure OpenAI AZURE_OPENAI_API_KEY=... AZURE_OPENAI_ENDPOINT=https://...openai.azure.com/ # Google Gemini GEMINI_API_KEY=... # MiniMax MINIMAX_API_KEY=... # Qwen QWEN_API_KEY=... ``` When these keys are present at runtime, the corresponding provider appears as selectable in [`components/settings/provider-config-panel.tsx`](https://github.com/THU-MAIC/OpenMAIC/blob/main/components/settings/provider-config-panel.tsx). ## Programmatic Provider Selection Switch between LLM providers programmatically using the settings store. The `setProvider()` and `setModel()` methods update the global state and persist preferences. ```typescript import { useSettingsStore } from '@/lib/store/settings'; // Switch to OpenAI GPT-4o useSettingsStore.getState().setProvider('openai'); useSettingsStore.getState().setModel('openai', 'gpt-4o'); // Switch to MiniMax M3 useSettingsStore.getState().setProvider('minimax'); useSettingsStore.getState().setModel('minimax', 'minimax-m3'); // Use local Lemonade for testing (no API key required) useSettingsStore.getState().setProvider('lemonade'); useSettingsStore.getState().setModel('lemonade', 'gpt-4o-mini'); ``` The store validates each selection against `providersConfig` to ensure the requested model exists for the selected provider ID. ## Adding Custom Providers Extend OpenMAIC to support self-hosted or niche OpenAI-compatible endpoints by mutating the `providersConfig` object at runtime. ```typescript import { useSettingsStore } from '@/lib/store/settings'; useSettingsStore.setState((state) => { state.providersConfig = { ...state.providersConfig, customLocalLLM: { id: 'customLocalLLM', type: 'openai', baseUrl: 'https://localhost:8080/v1', requiresApiKey: true, apiKey: process.env.CUSTOM_API_KEY, models: [ { id: 'llama-3-8b' }, { id: 'llama-3-70b' } ], }, }; }); ``` Custom providers must specify `type: 'openai'` to use the OpenAI-compatible request formatting implemented in [`lib/ai/providers.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/lib/ai/providers.ts). ## Validation and Testing The [`tests/store/settings-validation.test.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/tests/store/settings-validation.test.ts) file contains unit tests that verify provider configuration logic. These tests ensure that `hasUsableLLMProvider()` correctly identifies when credentials are missing and that `isLLMProviderConfigured()` returns accurate boolean states for each supported LLM provider for OpenMAIC. When adding new providers, extending these tests validates that the provider integrates correctly with the settings store and AI abstraction layer. ## Summary - OpenMAIC supports **eight LLM providers** natively: OpenAI, Anthropic, Azure OpenAI, Google Gemini, MiniMax, Qwen, Lemonade, and custom endpoints. - Provider configuration lives in **[`lib/store/settings.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/lib/store/settings.ts)** within the `providersConfig` object, validated by `hasUsableLLMProvider()` and `isLLMProviderConfigured()`. - Environment variables defined in **`.env.example`** activate specific providers at runtime. - **Lemonade** provides a local, keyless option for development environments. - Custom OpenAI-compatible APIs can be added by extending `providersConfig` with a unique provider ID and base URL. ## Frequently Asked Questions ### How do I switch between OpenAI and Anthropic models in OpenMAIC? Call `useSettingsStore.getState().setProvider('anthropic')` followed by `setModel('anthropic', 'claude-3-opus-20240229')`. The UI in [`components/settings/provider-config-panel.tsx`](https://github.com/THU-MAIC/OpenMAIC/blob/main/components/settings/provider-config-panel.tsx) reflects this change immediately, routing subsequent completions through Anthropic's API endpoint. ### Can I use OpenMAIC with a local LLM like Ollama or LM Studio? Yes. Add a custom provider entry with `type: 'openai'` and point `baseUrl` to your local OpenAI-compatible server (e.g., `http://localhost:11434/v1` for Ollama's compatibility mode). OpenMAIC will treat it like a standard cloud provider. ### What is the Lemonade provider used for? **Lemonade** is a built-in local mock server provider that mimics OpenAI's API structure without requiring API keys or internet connectivity. It is ideal for automated testing, offline development, and CI/CD pipelines where external API costs must be avoided. ### Where does OpenMAIC validate that my API keys are working? The [`tests/store/settings-validation.test.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/tests/store/settings-validation.test.ts) file contains the validation suite, while runtime checks occur in [`lib/store/settings.ts`](https://github.com/THU-MAIC/OpenMAIC/blob/main/lib/store/settings.ts) via `isLLMProviderConfigured()`. These functions verify that API keys are non-empty strings and that required base URLs are defined before allowing the provider to be selected in the interface.