Run China and global AI models in your self-hosted Open WebUI chat interface through one OpenAI-compatible endpoint.
Give a small team one self-hosted chat UI with shared access to China and global model families.
Compare DeepSeek, Qwen, GLM, GPT, Claude, Gemini, Kimi, and Doubao without creating separate provider accounts.
Use cheaper models for routine questions and stronger models for complex escalation or long documents.
API Base URL: https://api.aipower.me/v1
API Key: your-aipower-key
OPENAI_API_BASE_URL=https://api.aipower.me/v1 OPENAI_API_KEY=your-aipower-key
Open WebUI supports OpenAI-compatible provider connections. AIPower supplies the endpoint, model access, usage logs, and billing layer.
| Chat job | AIPower model | Why |
|---|---|---|
| General chat | deepseek/deepseek-chat | Good starter model for daily Q&A, drafts, and support-style replies. |
| Chinese chat | qwen/qwen-plus | Strong Chinese responses for local teams, customer notes, and bilingual workflows. |
| Fast classification | zhipu/glm-4-flash | Low-cost routing, tagging, short answers, and structured outputs. |
| Code and analysis | anthropic/claude-sonnet | Use after top-up for longer reasoning, coding, and technical analysis. |
| Long context | google/gemini-2.5-flash | Useful for long documents, policies, transcripts, and large pasted context. |
Models do not show up: check that your Base URL ends with /v1 and your AIPower key is active.
Premium models fail: first use trial-friendly models such as DeepSeek, Qwen, or GLM. GPT, Claude, Gemini, Kimi, and other premium routes may require top-up.
Team usage grows: move to Team/BYOK when you need shared budgets, provider-key routing, invoices, or accounting exports.
10 trial calls. One OpenAI-compatible endpoint. China + global models.
Get API key